chore: migrate 10 scalar operators to SQLGlot (#1922)

This change contains three commits generated by Gemini CLI tool:

- Migrated cos, hash, isnull, notnull, and sin operators.
- Migrated tan_op, arcsin_op, arccos_op, arctan_op, and sinh_op scalar operators to SQLGlot.

Fixes internal issue 430133370

Author: chelsea-lin

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

@@ -23,13 +23,52 @@
 from bigframes.core.compile.sqlglot.expressions.op_registration import OpRegistration
 from bigframes.core.compile.sqlglot.expressions.typed_expr import TypedExpr
 
+_NAN = sge.Cast(this=sge.convert("NaN"), to="FLOAT64")
+_INF = sge.Cast(this=sge.convert("Infinity"), to="FLOAT64")
+
+# Approx Highest number you can pass in to EXP function and get a valid FLOAT64 result
+# FLOAT64 has 11 exponent bits, so max values is about 2**(2**10)
+# ln(2**(2**10)) == (2**10)*ln(2) ~= 709.78, so EXP(x) for x>709.78 will overflow.
+_FLOAT64_EXP_BOUND = sge.convert(709.78)
+
 UNARY_OP_REGISTRATION = OpRegistration()
 
 
 def compile(op: ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
     return UNARY_OP_REGISTRATION[op](op, expr)
 
 
+@UNARY_OP_REGISTRATION.register(ops.arccos_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Case(
+        ifs=[
+            sge.If(
+                this=sge.func("ABS", expr.expr) > sge.convert(1),
+                true=_NAN,
+            )
+        ],
+        default=sge.func("ACOS", expr.expr),
+    )
+
+
+@UNARY_OP_REGISTRATION.register(ops.arcsin_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Case(
+        ifs=[
+            sge.If(
+                this=sge.func("ABS", expr.expr) > sge.convert(1),
+                true=_NAN,
+            )
+        ],
+        default=sge.func("ASIN", expr.expr),
+    )
+
+
+@UNARY_OP_REGISTRATION.register(ops.arctan_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func("ATAN", expr.expr)
+
+
 @UNARY_OP_REGISTRATION.register(ops.ArrayToStringOp)
 def _(op: ops.ArrayToStringOp, expr: TypedExpr) -> sge.Expression:
     return sge.ArrayToString(this=expr.expr, expression=f"'{op.delimiter}'")
@@ -72,6 +111,49 @@ def _(op: ops.ArraySliceOp, expr: TypedExpr) -> sge.Expression:
     return sge.array(selected_elements)
 
 
+@UNARY_OP_REGISTRATION.register(ops.cos_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func("COS", 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)
+
+
+@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())
+
+
+@UNARY_OP_REGISTRATION.register(ops.notnull_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Not(this=sge.Is(this=expr.expr, expression=sge.Null()))
+
+
+@UNARY_OP_REGISTRATION.register(ops.sin_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func("SIN", expr.expr)
+
+
+@UNARY_OP_REGISTRATION.register(ops.sinh_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.Case(
+        ifs=[
+            sge.If(
+                this=sge.func("ABS", expr.expr) > _FLOAT64_EXP_BOUND,
+                true=sge.func("SIGN", expr.expr) * _INF,
+            )
+        ],
+        default=sge.func("SINH", expr.expr),
+    )
+
+
+@UNARY_OP_REGISTRATION.register(ops.tan_op)
+def _(op: ops.base_ops.UnaryOp, expr: TypedExpr) -> sge.Expression:
+    return sge.func("TAN", expr.expr)
+
+
 # JSON Ops
 @UNARY_OP_REGISTRATION.register(ops.JSONExtract)
 def _(op: ops.JSONExtract, expr: TypedExpr) -> sge.Expression:

tests/unit/core/compile/sqlglot/expressions/snapshots/test_unary_compiler/test_arccos/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
+    *,
+    CASE WHEN ABS(`bfcol_0`) > 1 THEN CAST('NaN' AS FLOAT64) ELSE ACOS(`bfcol_0`) END 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_arcsin/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
+    *,
+    CASE WHEN ABS(`bfcol_0`) > 1 THEN CAST('NaN' AS FLOAT64) ELSE ASIN(`bfcol_0`) END 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_arctan/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
+    *,
+    ATAN(`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_cos/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
+    *,
+    COS(`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_hash/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
+    *,
+    FARM_FINGERPRINT(`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_isnull/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
+    *,
+    `bfcol_0` IS NULL 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_notnull/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
+    *,
+    NOT `bfcol_0` IS NULL 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_sin/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
+    *,
+    SIN(`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_sinh/out.sql

@@ -0,0 +1,17 @@
+WITH `bfcte_0` AS (
+  SELECT
+    `float64_col` AS `bfcol_0`
+  FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
+), `bfcte_1` AS (
+  SELECT
+    *,
+    CASE
+      WHEN ABS(`bfcol_0`) > 709.78
+      THEN SIGN(`bfcol_0`) * CAST('Infinity' AS FLOAT64)
+      ELSE SINH(`bfcol_0`)
+    END AS `bfcol_1`
+  FROM `bfcte_0`
+)
+SELECT
+  `bfcol_1` AS `float64_col`
+FROM `bfcte_1`