lit lit lit

Author: colin-ho

daft/expressions/expressions.py

@@ -110,7 +110,11 @@ def lit(value: object) -> Expression:
         assert isinstance(exponent, int)
         lit_value = _decimal_lit(sign == 1, digits, exponent)
     elif isinstance(value, Series):
-        lit_value = _series_lit(value._series)
+        series = Series.from_pylist([value]).cast(DataType.fixed_size_list(value.datatype(), len(value)))
+        lit_value = _series_lit(series._series)
+    elif isinstance(value, list):
+        list_of_exprs = [Expression._to_expression(item) for item in value]
+        lit_value = list_(*list_of_exprs)._expr
     else:
         lit_value = _lit(value)
     return Expression._from_pyexpr(lit_value)
@@ -1457,7 +1461,7 @@ def is_in(self, other: Any) -> Expression:
             other = [Expression._to_expression(item) for item in other]
         elif not isinstance(other, Expression):
             series = item_to_series("items", other)
-            other = [Expression._to_expression(series)]
+            other = [Expression._from_pyexpr(_series_lit(series._series))]
         else:
             other = [other]
 
@@ -4339,7 +4343,7 @@ def count_matches(
             patterns = [patterns]
         if not isinstance(patterns, Expression):
             series = item_to_series("items", patterns)
-            patterns = Expression._to_expression(series)
+            patterns = Expression._from_pyexpr(_series_lit(series._series))
 
         whole_words_expr = Expression._to_expression(whole_words)._expr
         case_sensitive_expr = Expression._to_expression(case_sensitive)._expr

tests/expressions/test_expressions.py

@@ -25,7 +25,7 @@
         (b"a", DataType.binary()),
         (True, DataType.bool()),
         (None, DataType.null()),
-        (Series.from_pylist([1, 2, 3]), DataType.int64()),
+        (Series.from_pylist([1, 2, 3]), DataType.fixed_size_list(DataType.int64(), 3)),
         (date(2023, 1, 1), DataType.date()),
         (time(1, 2, 3, 4), DataType.time(timeunit=TimeUnit.from_str("us"))),
         (datetime(2023, 1, 1), DataType.timestamp(timeunit=TimeUnit.from_str("us"))),
@@ -641,7 +641,7 @@ def test_duration_lit(input, expected) -> None:
 def test_repr_series_lit() -> None:
     s = lit(Series.from_pylist([1, 2, 3]))
     output = repr(s)
-    assert output == "lit([1, 2, 3])"
+    assert output == "lit([[1, 2, 3]])"
 
 
 def test_list_value_counts():

tests/expressions/test_null_safe_equals.py

@@ -3,7 +3,9 @@
 import pyarrow as pa
 import pytest
 
+from daft.daft import series_lit
 from daft.expressions import col, lit
+from daft.expressions.expressions import Expression
 from daft.recordbatch import MicroPartition
 
 
@@ -356,7 +358,13 @@ def test_length_mismatch_all_types(type_name, left_data, right_data):
     right_table = MicroPartition.from_pydict({"value": right_data})
 
     with pytest.raises(ValueError) as exc_info:
-        result = left_table.eval_expression_list([col("value").eq_null_safe(right_table.get_column_by_name("value"))])
+        result = left_table.eval_expression_list(
+            [
+                col("value").eq_null_safe(
+                    Expression._from_pyexpr(series_lit(right_table.get_column_by_name("value")._series))
+                )
+            ]
+        )
         # Force evaluation by accessing the result
         result.get_column_by_name("value").to_pylist()
 

tests/expressions/typing/test_float.py

@@ -36,7 +36,7 @@ def test_fill_nan(binary_data_fixture):
     lhs, rhs = binary_data_fixture
     assert_typing_resolve_vs_runtime_behavior(
         data=binary_data_fixture,
-        expr=col(lhs.name()).float.fill_nan(rhs),
+        expr=col(lhs.name()).float.fill_nan(col(rhs.name())),
         run_kernel=lambda: lhs.float.fill_nan(rhs),
         resolvable=(
             lhs.datatype() in (DataType.float32(), DataType.float64())