chore: implement ge, gt, le, gt compilers

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

@@ -140,14 +140,37 @@ def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @BINARY_OP_REGISTRATION.register(ops.ge_op)
 def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    return sge.GTE(this=left.expr, expression=right.expr)
+    left_expr = _coerce_bool_to_int(left)
+    right_expr = _coerce_bool_to_int(right)
+    return sge.GTE(this=left_expr, expression=right_expr)
+
+
+@BINARY_OP_REGISTRATION.register(ops.gt_op)
+def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    left_expr = _coerce_bool_to_int(left)
+    right_expr = _coerce_bool_to_int(right)
+    return sge.GT(this=left_expr, expression=right_expr)
 
 
 @BINARY_OP_REGISTRATION.register(ops.JSONSet)
 def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
     return sge.func("JSON_SET", left.expr, sge.convert(op.json_path), right.expr)
 
 
+@BINARY_OP_REGISTRATION.register(ops.lt_op)
+def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    left_expr = _coerce_bool_to_int(left)
+    right_expr = _coerce_bool_to_int(right)
+    return sge.LT(this=left_expr, expression=right_expr)
+
+
+@BINARY_OP_REGISTRATION.register(ops.le_op)
+def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    left_expr = _coerce_bool_to_int(left)
+    right_expr = _coerce_bool_to_int(right)
+    return sge.LTE(this=left_expr, expression=right_expr)
+
+
 @BINARY_OP_REGISTRATION.register(ops.mul_op)
 def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
     left_expr = _coerce_bool_to_int(left)
@@ -170,6 +193,11 @@ def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
     return sge.NEQ(this=left_expr, expression=right_expr)
 
 
+@BINARY_OP_REGISTRATION.register(ops.obj_make_ref_op)
+def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    return sge.func("OBJ.MAKE_REF", left.expr, right.expr)
+
+
 @BINARY_OP_REGISTRATION.register(ops.sub_op)
 def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
     if dtypes.is_numeric(left.dtype) and dtypes.is_numeric(right.dtype):
@@ -202,11 +230,6 @@ def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
     )
 
 
-@BINARY_OP_REGISTRATION.register(ops.obj_make_ref_op)
-def _(op, left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    return sge.func("OBJ.MAKE_REF", left.expr, right.expr)
-
-
 def _coerce_bool_to_int(typed_expr: TypedExpr) -> sge.Expression:
     """Coerce boolean expression to integer."""
     if typed_expr.dtype == dtypes.BOOL_DTYPE:

tests/system/small/engines/test_comparison_ops.py

@@ -48,7 +48,7 @@ def apply_op_pairwise(
     return new_arr
 
 
-@pytest.mark.parametrize("engine", ["polars", "bq"], indirect=True)
+@pytest.mark.parametrize("engine", ["polars", "bq", "bq-sqlglot"], indirect=True)
 @pytest.mark.parametrize(
     "op",
     [

tests/unit/core/compile/sqlglot/expressions/snapshots/test_binary_compiler/test_ge_numeric/out.sql

@@ -0,0 +1,54 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `rowindex` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    `bfcol_2` AS `bfcol_6`,
+    `bfcol_1` AS `bfcol_7`,
+    `bfcol_0` AS `bfcol_8`,
+    `bfcol_1` >= `bfcol_1` AS `bfcol_9`
+  FROM `bfcte_0`
+), `bfcte_2` AS (
+  SELECT
+    *,
+    `bfcol_6` AS `bfcol_14`,
+    `bfcol_7` AS `bfcol_15`,
+    `bfcol_8` AS `bfcol_16`,
+    `bfcol_9` AS `bfcol_17`,
+    `bfcol_7` >= 1 AS `bfcol_18`
+  FROM `bfcte_1`
+), `bfcte_3` AS (
+  SELECT
+    *,
+    `bfcol_14` AS `bfcol_24`,
+    `bfcol_15` AS `bfcol_25`,
+    `bfcol_16` AS `bfcol_26`,
+    `bfcol_17` AS `bfcol_27`,
+    `bfcol_18` AS `bfcol_28`,
+    `bfcol_15` >= CAST(`bfcol_16` AS INT64) AS `bfcol_29`
+  FROM `bfcte_2`
+), `bfcte_4` AS (
+  SELECT
+    *,
+    `bfcol_24` AS `bfcol_36`,
+    `bfcol_25` AS `bfcol_37`,
+    `bfcol_26` AS `bfcol_38`,
+    `bfcol_27` AS `bfcol_39`,
+    `bfcol_28` AS `bfcol_40`,
+    `bfcol_29` AS `bfcol_41`,
+    CAST(`bfcol_26` AS INT64) >= `bfcol_25` AS `bfcol_42`
+  FROM `bfcte_3`
+)
+SELECT
+  `bfcol_36` AS `rowindex`,
+  `bfcol_37` AS `int64_col`,
+  `bfcol_38` AS `bool_col`,
+  `bfcol_39` AS `int_ge_int`,
+  `bfcol_40` AS `int_ge_1`,
+  `bfcol_41` AS `int_ge_bool`,
+  `bfcol_42` AS `bool_ge_int`
+FROM `bfcte_4`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_binary_compiler/test_gt_numeric/out.sql

@@ -0,0 +1,54 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `rowindex` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    `bfcol_2` AS `bfcol_6`,
+    `bfcol_1` AS `bfcol_7`,
+    `bfcol_0` AS `bfcol_8`,
+    `bfcol_1` > `bfcol_1` AS `bfcol_9`
+  FROM `bfcte_0`
+), `bfcte_2` AS (
+  SELECT
+    *,
+    `bfcol_6` AS `bfcol_14`,
+    `bfcol_7` AS `bfcol_15`,
+    `bfcol_8` AS `bfcol_16`,
+    `bfcol_9` AS `bfcol_17`,
+    `bfcol_7` > 1 AS `bfcol_18`
+  FROM `bfcte_1`
+), `bfcte_3` AS (
+  SELECT
+    *,
+    `bfcol_14` AS `bfcol_24`,
+    `bfcol_15` AS `bfcol_25`,
+    `bfcol_16` AS `bfcol_26`,
+    `bfcol_17` AS `bfcol_27`,
+    `bfcol_18` AS `bfcol_28`,
+    `bfcol_15` > CAST(`bfcol_16` AS INT64) AS `bfcol_29`
+  FROM `bfcte_2`
+), `bfcte_4` AS (
+  SELECT
+    *,
+    `bfcol_24` AS `bfcol_36`,
+    `bfcol_25` AS `bfcol_37`,
+    `bfcol_26` AS `bfcol_38`,
+    `bfcol_27` AS `bfcol_39`,
+    `bfcol_28` AS `bfcol_40`,
+    `bfcol_29` AS `bfcol_41`,
+    CAST(`bfcol_26` AS INT64) > `bfcol_25` AS `bfcol_42`
+  FROM `bfcte_3`
+)
+SELECT
+  `bfcol_36` AS `rowindex`,
+  `bfcol_37` AS `int64_col`,
+  `bfcol_38` AS `bool_col`,
+  `bfcol_39` AS `int_gt_int`,
+  `bfcol_40` AS `int_gt_1`,
+  `bfcol_41` AS `int_gt_bool`,
+  `bfcol_42` AS `bool_gt_int`
+FROM `bfcte_4`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_binary_compiler/test_le_numeric/out.sql

@@ -0,0 +1,54 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `rowindex` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    `bfcol_2` AS `bfcol_6`,
+    `bfcol_1` AS `bfcol_7`,
+    `bfcol_0` AS `bfcol_8`,
+    `bfcol_1` <= `bfcol_1` AS `bfcol_9`
+  FROM `bfcte_0`
+), `bfcte_2` AS (
+  SELECT
+    *,
+    `bfcol_6` AS `bfcol_14`,
+    `bfcol_7` AS `bfcol_15`,
+    `bfcol_8` AS `bfcol_16`,
+    `bfcol_9` AS `bfcol_17`,
+    `bfcol_7` <= 1 AS `bfcol_18`
+  FROM `bfcte_1`
+), `bfcte_3` AS (
+  SELECT
+    *,
+    `bfcol_14` AS `bfcol_24`,
+    `bfcol_15` AS `bfcol_25`,
+    `bfcol_16` AS `bfcol_26`,
+    `bfcol_17` AS `bfcol_27`,
+    `bfcol_18` AS `bfcol_28`,
+    `bfcol_15` <= CAST(`bfcol_16` AS INT64) AS `bfcol_29`
+  FROM `bfcte_2`
+), `bfcte_4` AS (
+  SELECT
+    *,
+    `bfcol_24` AS `bfcol_36`,
+    `bfcol_25` AS `bfcol_37`,
+    `bfcol_26` AS `bfcol_38`,
+    `bfcol_27` AS `bfcol_39`,
+    `bfcol_28` AS `bfcol_40`,
+    `bfcol_29` AS `bfcol_41`,
+    CAST(`bfcol_26` AS INT64) <= `bfcol_25` AS `bfcol_42`
+  FROM `bfcte_3`
+)
+SELECT
+  `bfcol_36` AS `rowindex`,
+  `bfcol_37` AS `int64_col`,
+  `bfcol_38` AS `bool_col`,
+  `bfcol_39` AS `int_le_int`,
+  `bfcol_40` AS `int_le_1`,
+  `bfcol_41` AS `int_le_bool`,
+  `bfcol_42` AS `bool_le_int`
+FROM `bfcte_4`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_binary_compiler/test_lt_numeric/out.sql

@@ -0,0 +1,54 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `bool_col` AS `bfcol_0`,
+    `int64_col` AS `bfcol_1`,
+    `rowindex` AS `bfcol_2`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    `bfcol_2` AS `bfcol_6`,
+    `bfcol_1` AS `bfcol_7`,
+    `bfcol_0` AS `bfcol_8`,
+    `bfcol_1` < `bfcol_1` AS `bfcol_9`
+  FROM `bfcte_0`
+), `bfcte_2` AS (
+  SELECT
+    *,
+    `bfcol_6` AS `bfcol_14`,
+    `bfcol_7` AS `bfcol_15`,
+    `bfcol_8` AS `bfcol_16`,
+    `bfcol_9` AS `bfcol_17`,
+    `bfcol_7` < 1 AS `bfcol_18`
+  FROM `bfcte_1`
+), `bfcte_3` AS (
+  SELECT
+    *,
+    `bfcol_14` AS `bfcol_24`,
+    `bfcol_15` AS `bfcol_25`,
+    `bfcol_16` AS `bfcol_26`,
+    `bfcol_17` AS `bfcol_27`,
+    `bfcol_18` AS `bfcol_28`,
+    `bfcol_15` < CAST(`bfcol_16` AS INT64) AS `bfcol_29`
+  FROM `bfcte_2`
+), `bfcte_4` AS (
+  SELECT
+    *,
+    `bfcol_24` AS `bfcol_36`,
+    `bfcol_25` AS `bfcol_37`,
+    `bfcol_26` AS `bfcol_38`,
+    `bfcol_27` AS `bfcol_39`,
+    `bfcol_28` AS `bfcol_40`,
+    `bfcol_29` AS `bfcol_41`,
+    CAST(`bfcol_26` AS INT64) < `bfcol_25` AS `bfcol_42`
+  FROM `bfcte_3`
+)
+SELECT
+  `bfcol_36` AS `rowindex`,
+  `bfcol_37` AS `int64_col`,
+  `bfcol_38` AS `bool_col`,
+  `bfcol_39` AS `int_lt_int`,
+  `bfcol_40` AS `int_lt_1`,
+  `bfcol_41` AS `int_lt_bool`,
+  `bfcol_42` AS `bool_lt_int`
+FROM `bfcte_4`

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

@@ -149,6 +149,30 @@ def test_floordiv_timedelta(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(bf_df.sql, "out.sql")
 
 
+def test_gt_numeric(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "bool_col"]]
+
+    bf_df["int_gt_int"] = bf_df["int64_col"] > bf_df["int64_col"]
+    bf_df["int_gt_1"] = bf_df["int64_col"] > 1
+
+    bf_df["int_gt_bool"] = bf_df["int64_col"] > bf_df["bool_col"]
+    bf_df["bool_gt_int"] = bf_df["bool_col"] > bf_df["int64_col"]
+
+    snapshot.assert_match(bf_df.sql, "out.sql")
+
+
+def test_ge_numeric(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "bool_col"]]
+
+    bf_df["int_ge_int"] = bf_df["int64_col"] >= bf_df["int64_col"]
+    bf_df["int_ge_1"] = bf_df["int64_col"] >= 1
+
+    bf_df["int_ge_bool"] = bf_df["int64_col"] >= bf_df["bool_col"]
+    bf_df["bool_ge_int"] = bf_df["bool_col"] >= bf_df["int64_col"]
+
+    snapshot.assert_match(bf_df.sql, "out.sql")
+
+
 def test_json_set(json_types_df: bpd.DataFrame, snapshot):
     bf_df = json_types_df[["json_col"]]
     sql = _apply_binary_op(
@@ -158,6 +182,30 @@ def test_json_set(json_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_lt_numeric(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "bool_col"]]
+
+    bf_df["int_lt_int"] = bf_df["int64_col"] < bf_df["int64_col"]
+    bf_df["int_lt_1"] = bf_df["int64_col"] < 1
+
+    bf_df["int_lt_bool"] = bf_df["int64_col"] < bf_df["bool_col"]
+    bf_df["bool_lt_int"] = bf_df["bool_col"] < bf_df["int64_col"]
+
+    snapshot.assert_match(bf_df.sql, "out.sql")
+
+
+def test_le_numeric(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col", "bool_col"]]
+
+    bf_df["int_le_int"] = bf_df["int64_col"] <= bf_df["int64_col"]
+    bf_df["int_le_1"] = bf_df["int64_col"] <= 1
+
+    bf_df["int_le_bool"] = bf_df["int64_col"] <= bf_df["bool_col"]
+    bf_df["bool_le_int"] = bf_df["bool_col"] <= bf_df["int64_col"]
+
+    snapshot.assert_match(bf_df.sql, "out.sql")
+
+
 def test_sub_numeric(scalar_types_df: bpd.DataFrame, snapshot):
     bf_df = scalar_types_df[["int64_col", "bool_col"]]