chore: Migrate cosine_distance_op operator to SQLGlot (#2236)

Fixes b/447388852 🦕

Author: jialuoo

bigframes/core/compile/sqlglot/expressions/numeric_ops.py

@@ -125,6 +125,18 @@ def _(expr: TypedExpr) -> sge.Expression:
     )
 
 
+@register_binary_op(ops.cosine_distance_op)
+def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    return sge.Anonymous(
+        this="ML.DISTANCE",
+        expressions=[
+            left.expr,
+            right.expr,
+            sge.Literal.string("COSINE"),
+        ],
+    )
+
+
 @register_unary_op(ops.exp_op)
 def _(expr: TypedExpr) -> sge.Expression:
     return sge.Case(

tests/unit/core/compile/sqlglot/expressions/snapshots/test_numeric_ops/test_cosine_distance/out.sql

@@ -0,0 +1,16 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `int_list_col` AS `bfcol_0`,
+    `float_list_col` AS `bfcol_1`
+  FROM `bigframes-dev`.`sqlglot_test`.`repeated_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    ML.DISTANCE(`bfcol_0`, `bfcol_0`, 'COSINE') AS `bfcol_2`,
+    ML.DISTANCE(`bfcol_1`, `bfcol_1`, 'COSINE') AS `bfcol_3`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_2` AS `int_list_col`,
+  `bfcol_3` AS `float_list_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/test_numeric_ops.py

@@ -117,6 +117,21 @@ def test_cosh(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_cosine_distance(repeated_types_df: bpd.DataFrame, snapshot):
+    col_names = ["int_list_col", "float_list_col"]
+    bf_df = repeated_types_df[col_names]
+
+    sql = utils._apply_ops_to_sql(
+        bf_df,
+        [
+            ops.cosine_distance_op.as_expr("int_list_col", "int_list_col"),
+            ops.cosine_distance_op.as_expr("float_list_col", "float_list_col"),
+        ],
+        ["int_list_col", "float_list_col"],
+    )
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_exp(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "float64_col"
     bf_df = scalar_types_df[[col_name]]