chore: Migrate up to 5 scalar operators to SQLGlot

Migrated operators:
- abs_op
- capitalize_op
- ceil_op
- date_op
- day_op

Author: chelsea-lin

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

@@ -38,6 +38,11 @@ def compile(op: ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return UNARY_OP_REGISTRATION[op](op, expr)
 
 
+@UNARY_OP_REGISTRATION.register(ops.abs_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Abs(this=expr.expr)
+
+
 @UNARY_OP_REGISTRATION.register(ops.arccosh_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.Case(
@@ -142,6 +147,16 @@ def _(op: ops.ArraySliceOp, expr: TypedExpr) -> sge.Expression:
     return sge.array(selected_elements)
 
 
+@UNARY_OP_REGISTRATION.register(ops.capitalize_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Initcap(this=expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.ceil_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Ceil(this=expr.expr)
+
+
 @UNARY_OP_REGISTRATION.register(ops.cos_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.func("COS", expr.expr)
@@ -160,6 +175,16 @@ def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     )
 
 
+@UNARY_OP_REGISTRATION.register(ops.date_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Date(this=expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.day_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Extract(this=sge.Identifier(this="DAY"), expression=expr.expr)
+
+
 @UNARY_OP_REGISTRATION.register(ops.hash_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.func("FARM_FINGERPRINT", expr.expr)

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_abs/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `float64_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    ABS(`bfcol_0`) AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `float64_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_capitalize/out.sql

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

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_ceil/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `float64_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    CEIL(`bfcol_0`) AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `float64_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_date/out.sql

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

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_day/out.sql

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

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

@@ -73,6 +73,36 @@ def test_arctanh(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_abs(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["float64_col"]]
+    sql = _apply_unary_op(bf_df, ops.abs_op, "float64_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_capitalize(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.capitalize_op, "string_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_ceil(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["float64_col"]]
+    sql = _apply_unary_op(bf_df, ops.ceil_op, "float64_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_date(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["timestamp_col"]]
+    sql = _apply_unary_op(bf_df, ops.date_op, "timestamp_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_day(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["timestamp_col"]]
+    sql = _apply_unary_op(bf_df, ops.day_op, "timestamp_col")
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_array_to_string(repeated_types_df: bpd.DataFrame, snapshot):
     bf_df = repeated_types_df[["string_list_col"]]
     sql = _apply_unary_op(bf_df, ops.ArrayToStringOp(delimiter="."), "string_list_col")