Merge branch 'main' into b455813915-pyformat-str

Author: sycai

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

@@ -21,6 +21,7 @@
 import bigframes.core.compile.sqlglot.scalar_compiler as scalar_compiler
 
 register_unary_op = scalar_compiler.scalar_op_compiler.register_unary_op
+register_binary_op = scalar_compiler.scalar_op_compiler.register_binary_op
 
 
 @register_unary_op(ops.geo_area_op)
@@ -108,3 +109,8 @@ def _(expr: TypedExpr) -> sge.Expression:
 @register_unary_op(ops.geo_y_op)
 def _(expr: TypedExpr) -> sge.Expression:
     return sge.func("SAFE.ST_Y", expr.expr)
+
+
+@register_binary_op(ops.geo_st_difference_op)
+def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    return sge.func("ST_DIFFERENCE", left.expr, right.expr)

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

@@ -77,6 +77,13 @@ def _(expr: TypedExpr) -> sge.Expression:
     return sge.func("ASINH", expr.expr)
 
 
+@register_binary_op(ops.arctan2_op)
+def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    left_expr = _coerce_bool_to_int(left)
+    right_expr = _coerce_bool_to_int(right)
+    return sge.func("ATAN2", left_expr, right_expr)
+
+
 @register_unary_op(ops.arctan_op)
 def _(expr: TypedExpr) -> sge.Expression:
     return sge.func("ATAN", expr.expr)
@@ -118,6 +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_geo_ops/test_geo_st_difference/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `geography_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    ST_DIFFERENCE(`bfcol_0`, `bfcol_0`) AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `geography_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_numeric_ops/test_arctan2/out.sql

@@ -0,0 +1,17 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `float64_col` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    ATAN2(`bfcol_1`, `bfcol_2`) AS `bfcol_6`,
+    ATAN2(CAST(`bfcol_0` AS INT64), `bfcol_2`) AS `bfcol_7`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_6` AS `int64_col`,
+  `bfcol_7` AS `bool_col`
+FROM `bfcte_1`

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_geo_ops.py

@@ -81,6 +81,14 @@ def test_geo_st_convexhull(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_geo_st_difference(scalar_types_df: bpd.DataFrame, snapshot):
+    col_name = "geography_col"
+    bf_df = scalar_types_df[[col_name]]
+    sql = utils._apply_binary_op(bf_df, ops.geo_st_difference_op, col_name, col_name)
+
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_geo_st_geogfromtext(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "string_col"
     bf_df = scalar_types_df[[col_name]]

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

@@ -55,6 +55,20 @@ def test_arcsinh(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_arctan2(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "float64_col", "bool_col"]]
+
+    sql = utils._apply_ops_to_sql(
+        bf_df,
+        [
+            ops.arctan2_op.as_expr("int64_col", "float64_col"),
+            ops.arctan2_op.as_expr("bool_col", "float64_col"),
+        ],
+        ["int64_col", "bool_col"],
+    )
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_arctan(scalar_types_df: bpd.DataFrame, snapshot):
     col_name = "float64_col"
     bf_df = scalar_types_df[[col_name]]
@@ -103,6 +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]]