chore: migrate 12 scalar operators to SQLGlot

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

@@ -234,6 +234,84 @@ def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.func("FARM_FINGERPRINT", expr.expr)
 
 
+@UNARY_OP_REGISTRATION.register(ops.hour_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Extract(this=sge.Identifier(this="HOUR"), expression=expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.invert_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.BitwiseNot(this=expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.isalnum_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^(\p{N}|\p{L})+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.isalpha_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^\p{L}+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.isdecimal_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^\d+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.isdigit_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^\p{Nd}+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.islower_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.And(
+        this=sge.EQ(
+            this=sge.Lower(this=expr.expr),
+            expression=expr.expr,
+        ),
+        expression=sge.NEQ(
+            this=sge.Upper(this=expr.expr),
+            expression=expr.expr,
+        ),
+    )
+
+
+@UNARY_OP_REGISTRATION.register(ops.isnumeric_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^\pN+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.isspace_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.RegexpLike(this=expr.expr, expression=sge.convert(r"^\s+$"))
+
+
+@UNARY_OP_REGISTRATION.register(ops.isupper_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.And(
+        this=sge.EQ(
+            this=sge.Upper(this=expr.expr),
+            expression=expr.expr,
+        ),
+        expression=sge.NEQ(
+            this=sge.Lower(this=expr.expr),
+            expression=expr.expr,
+        ),
+    )
+
+
+@UNARY_OP_REGISTRATION.register(ops.iso_day_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Extract(this=sge.Identifier(this="DAYOFWEEK"), expression=expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.iso_week_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Extract(this=sge.Identifier(this="ISOWEEK"), expression=expr.expr)
+
+
 @UNARY_OP_REGISTRATION.register(ops.isnull_op)
 def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return sge.Is(this=expr.expr, expression=sge.Null())

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_hour/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(HOUR FROM `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_invert/out.sql

@@ -0,0 +1,13 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `int64_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    ~`bfcol_0` AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `int64_col`
+FROM `bfcte_1`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_isalnum/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^(\\p{N}|\\p{L})+$') 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_isalpha/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^\\p{L}+$') 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_isdecimal/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^\\d+$') 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_isdigit/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^\\p{Nd}+$') 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_islower/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
+    *,
+    LOWER(`bfcol_0`) = `bfcol_0` AND UPPER(`bfcol_0`) <> `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_isnumeric/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^\\pN+$') 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_iso_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(DAYOFWEEK FROM `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_iso_week/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(ISOWEEK FROM `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_isspace/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
+    *,
+    REGEXP_CONTAINS(`bfcol_0`, '^\\s+$') 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_isupper/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
+    *,
+    UPPER(`bfcol_0`) = `bfcol_0` AND LOWER(`bfcol_0`) <> `bfcol_0` AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `string_col`
+FROM `bfcte_1`

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

@@ -93,6 +93,7 @@ def test_capitalize(scalar_types_df: bpd.DataFrame, snapshot):
 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")
 
 
@@ -194,6 +195,90 @@ def test_hash(scalar_types_df: bpd.DataFrame, snapshot):
     snapshot.assert_match(sql, "out.sql")
 
 
+def test_hour(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["timestamp_col"]]
+    sql = _apply_unary_op(bf_df, ops.hour_op, "timestamp_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_invert(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["int64_col"]]
+    sql = _apply_unary_op(bf_df, ops.invert_op, "int64_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isalnum(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isalnum_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isalpha(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isalpha_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isdecimal(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isdecimal_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isdigit(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isdigit_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_islower(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.islower_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isnumeric(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isnumeric_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isspace(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isspace_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_isupper(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["string_col"]]
+    sql = _apply_unary_op(bf_df, ops.isupper_op, "string_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_iso_day(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["timestamp_col"]]
+    sql = _apply_unary_op(bf_df, ops.iso_day_op, "timestamp_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
+def test_iso_week(scalar_types_df: bpd.DataFrame, snapshot):
+    bf_df = scalar_types_df[["timestamp_col"]]
+    sql = _apply_unary_op(bf_df, ops.iso_week_op, "timestamp_col")
+
+    snapshot.assert_match(sql, "out.sql")
+
+
 def test_isnull(scalar_types_df: bpd.DataFrame, snapshot):
     bf_df = scalar_types_df[["float64_col"]]
     sql = _apply_unary_op(bf_df, ops.isnull_op, "float64_col")