refactor: add agg_ops.MeanOp for sqlglot compiler

Author: chelsea-lin

bigframes/core/compile/sqlglot/aggregations/unary_compiler.py

@@ -55,6 +55,26 @@ def _(
     return apply_window_if_present(sge.func("MAX", column.expr), window)
 
 
+@UNARY_OP_REGISTRATION.register(agg_ops.MeanOp)
+def _(
+    op: agg_ops.MeanOp,
+    column: typed_expr.TypedExpr,
+    window: typing.Optional[window_spec.WindowSpec] = None,
+) -> sge.Expression:
+    expr = column.expr
+    if column.dtype == dtypes.BOOL_DTYPE:
+        expr = sge.Cast(this=expr, to="INT64")
+
+    expr = sge.func("AVG", expr)
+
+    should_floor_result = (
+        op.should_floor_result or column.dtype == dtypes.TIMEDELTA_DTYPE
+    )
+    if should_floor_result:
+        expr = sge.func("FLOOR", expr)
+    return apply_window_if_present(expr, window)
+
+
 @UNARY_OP_REGISTRATION.register(agg_ops.MinOp)
 def _(
     op: agg_ops.MinOp,

tests/system/small/engines/test_aggregation.py

@@ -70,7 +70,7 @@ def test_engines_aggregate_size(
     assert_equivalence_execution(node, REFERENCE_ENGINE, engine)
 
 
-@pytest.mark.parametrize("engine", ["polars", "bq"], indirect=True)
+@pytest.mark.parametrize("engine", ["polars", "bq", "bq-sqlglot"], indirect=True)
 @pytest.mark.parametrize(
     "op",
     [agg_ops.min_op, agg_ops.max_op, agg_ops.mean_op, agg_ops.sum_op, agg_ops.count_op],

tests/unit/core/compile/sqlglot/aggregations/snapshots/test_unary_compiler/test_mean/out.sql

@@ -0,0 +1,20 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `duration_col` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    AVG(`bfcol_1`) AS `bfcol_6`,
+    AVG(CAST(`bfcol_0` AS INT64)) AS `bfcol_7`,
+    AVG(`bfcol_2`) AS `bfcol_8`,
+    FLOOR(AVG(`bfcol_1`)) AS `bfcol_9`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_6` AS `int64_col`,
+  `bfcol_7` AS `bool_col`,
+  `bfcol_8` AS `duration_col`,
+  `bfcol_9` AS `int64_col_w_floor`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/aggregations/test_unary_compiler.py

@@ -56,6 +56,25 @@ def test_max(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_mean(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "bool_col", "duration_col"]]
+    # bf_df["duration_col"] = bpd.to_timedelta(bf_df["duration_col"], unit="us")
+
+    agg_ops_map = {
+        "int64_col": agg_ops.MeanOp().as_expr("int64_col"),
+        "bool_col": agg_ops.MeanOp().as_expr("bool_col"),
+        "duration_col": agg_ops.MeanOp().as_expr("duration_col"),
+        "int64_col_w_floor": agg_ops.MeanOp(should_floor_result=True).as_expr(
+            "int64_col"
+        ),
+    }
+    sql = _apply_unary_agg_ops(
+        bf_df, list(agg_ops_map.values()), list(agg_ops_map.keys())
+    )
+
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_min(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "int64_col"
     bf_df = scalar_types_df[[col_name]]