Merge branch 'main' into shuowei-job-history

Author: shuoweil

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

@@ -18,6 +18,7 @@
 
 from bigframes import dtypes
 from bigframes import operations as ops
+from bigframes.core.compile.sqlglot import sql
 import bigframes.core.compile.sqlglot.expression_compiler as expression_compiler
 from bigframes.core.compile.sqlglot.expressions.typed_expr import TypedExpr
 
@@ -29,10 +30,10 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
     # For AND, when we encounter a NULL value, we only know when the result is FALSE,
     # otherwise the result is unknown (NULL). See: truth table at
     # https://en.wikibooks.org/wiki/Structured_Query_Language/NULLs_and_the_Three_Valued_Logic#AND,_OR
-    if left.expr == sge.null():
+    if sql.is_null_literal(left.expr):
         condition = sge.EQ(this=right.expr, expression=sge.convert(False))
         return sge.If(this=condition, true=right.expr, false=sge.null())
-    if right.expr == sge.null():
+    if sql.is_null_literal(right.expr):
         condition = sge.EQ(this=left.expr, expression=sge.convert(False))
         return sge.If(this=condition, true=left.expr, false=sge.null())
 
@@ -46,10 +47,10 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
     # For OR, when we encounter a NULL value, we only know when the result is TRUE,
     # otherwise the result is unknown (NULL). See: truth table at
     # https://en.wikibooks.org/wiki/Structured_Query_Language/NULLs_and_the_Three_Valued_Logic#AND,_OR
-    if left.expr == sge.null():
+    if sql.is_null_literal(left.expr):
         condition = sge.EQ(this=right.expr, expression=sge.convert(True))
         return sge.If(this=condition, true=right.expr, false=sge.null())
-    if right.expr == sge.null():
+    if sql.is_null_literal(right.expr):
         condition = sge.EQ(this=left.expr, expression=sge.convert(True))
         return sge.If(this=condition, true=left.expr, false=sge.null())
 
@@ -64,12 +65,12 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
     # maintains the boolean data type.
     left_expr = left.expr
     left_dtype = left.dtype
-    if left_expr == sge.null():
+    if sql.is_null_literal(left_expr):
         left_expr = sge.Cast(this=sge.convert(None), to="BOOLEAN")
         left_dtype = dtypes.BOOL_DTYPE
     right_expr = right.expr
     right_dtype = right.dtype
-    if right_expr == sge.null():
+    if sql.is_null_literal(right_expr):
         right_expr = sge.Cast(this=sge.convert(None), to="BOOLEAN")
         right_dtype = dtypes.BOOL_DTYPE
 

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

@@ -102,7 +102,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.ge_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)
@@ -112,7 +112,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.gt_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)
@@ -122,7 +122,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.lt_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)
@@ -132,7 +132,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.le_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)

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

@@ -19,6 +19,7 @@
 
 from bigframes import dtypes
 from bigframes import operations as ops
+from bigframes.core.compile.sqlglot import sql
 import bigframes.core.compile.sqlglot.expression_compiler as expression_compiler
 from bigframes.core.compile.sqlglot.expressions.common import round_towards_zero
 import bigframes.core.compile.sqlglot.expressions.constants as constants
@@ -260,6 +261,9 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 def _int_pow_op(
     left_expr: sge.Expression, right_expr: sge.Expression
 ) -> sge.Expression:
+    if sql.is_null_literal(left_expr) or sql.is_null_literal(right_expr):
+        return sge.null()
+
     overflow_cond = sge.and_(
         sge.NEQ(this=left_expr, expression=sge.convert(0)),
         sge.GT(
@@ -292,6 +296,9 @@ def _int_pow_op(
 def _float_pow_op(
     left_expr: sge.Expression, right_expr: sge.Expression
 ) -> sge.Expression:
+    if sql.is_null_literal(left_expr) or sql.is_null_literal(right_expr):
+        return sge.null()
+
     # Most conditions here seek to prevent calling BQ POW with inputs that would generate errors.
     # See: https://cloud.google.com/bigquery/docs/reference/standard-sql/mathematical_functions#pow
     overflow_cond = sge.and_(
@@ -425,7 +432,7 @@ def _(expr: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.add_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     if left.dtype == dtypes.STRING_DTYPE and right.dtype == dtypes.STRING_DTYPE:
@@ -463,6 +470,9 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.div_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
+        return sge.null()
+
     left_expr = _coerce_bool_to_int(left)
     right_expr = _coerce_bool_to_int(right)
 
@@ -482,7 +492,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.floordiv_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)
@@ -525,6 +535,9 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.mod_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
+        return sge.null()
+
     # In BigQuery returned value has the same sign as X. In pandas, the sign of y is used, so we need to flip the result if sign(x) != sign(y)
     left_expr = _coerce_bool_to_int(left)
     right_expr = _coerce_bool_to_int(right)
@@ -568,7 +581,7 @@ def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.mul_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     left_expr = _coerce_bool_to_int(left)
@@ -594,7 +607,7 @@ def _(expr: TypedExpr, n_digits: TypedExpr) -> sge.Expression:
 
 @register_binary_op(ops.sub_op)
 def _(left: TypedExpr, right: TypedExpr) -> sge.Expression:
-    if left.expr == sge.null() or right.expr == sge.null():
+    if sql.is_null_literal(left.expr) or sql.is_null_literal(right.expr):
         return sge.null()
 
     if dtypes.is_numeric(left.dtype) and dtypes.is_numeric(right.dtype):

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

@@ -366,10 +366,16 @@ def string_slice(
                     column_length + sge.convert(start + 1),
                 ]
             )
-            length_expr = sge.convert(op_end) - sge.Greatest(
+            length_expr = sge.Greatest(
                 expressions=[
                     sge.convert(0),
-                    column_length + sge.convert(start),
+                    sge.convert(op_end)
+                    - sge.Greatest(
+                        expressions=[
+                            sge.convert(0),
+                            column_length + sge.convert(start),
+                        ]
+                    ),
                 ]
             )
         else:

tests/system/small/test_dataframe.py

@@ -6296,3 +6296,20 @@ def test_agg_with_dict_containing_non_existing_col_raise_key_error(scalars_dfs):
 
     with pytest.raises(KeyError):
         bf_df.agg(agg_funcs)
+
+
+def test_empty_agg_projection_succeeds():
+    # Tests that the compiler generates a SELECT 1 fallback for empty aggregations,
+    # protecting against BigQuery syntax errors when both groups and metrics are empty.
+    import importlib
+
+    bq = importlib.import_module(
+        "bigframes_vendored.ibis.backends.sql.compilers.bigquery"
+    )
+    sg = importlib.import_module("bigframes_vendored.sqlglot")
+
+    compiler = bq.BigQueryCompiler()
+    res = compiler.visit_Aggregate(
+        "op", parent=sg.table("parent_table"), groups=[], metrics=[]
+    )
+    assert "SELECT 1" in res.sql()

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

@@ -6,7 +6,7 @@ SELECT
   IEEE_DIVIDE(`int64_col`, `int64_col`) AS `int_div_int`,
   IEEE_DIVIDE(`int64_col`, 1) AS `int_div_1`,
   IEEE_DIVIDE(`int64_col`, 0.0) AS `int_div_0`,
-  IEEE_DIVIDE(`int64_col`, NULL) AS `int_div_null`,
+  NULL AS `int_div_null`,
   IEEE_DIVIDE(`int64_col`, `float64_col`) AS `int_div_float`,
   IEEE_DIVIDE(`float64_col`, `int64_col`) AS `float_div_int`,
   IEEE_DIVIDE(`float64_col`, 0.0) AS `float_div_0`,

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

@@ -34,6 +34,7 @@ SELECT
     THEN CAST('Infinity' AS FLOAT64) * `float64_col`
     ELSE CAST(FLOOR(IEEE_DIVIDE(`float64_col`, 0.0)) AS INT64)
   END AS `float_div_0`,
+  NULL AS `float_div_null`,
   CASE
     WHEN CAST(`bool_col` AS INT64) = CAST(0 AS INT64)
     THEN CAST(0 AS INT64) * `int64_col`

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

@@ -189,5 +189,6 @@ SELECT
       MOD(CAST(`float64_col` AS BIGNUMERIC), CAST(0 AS BIGNUMERIC))
     )
     ELSE MOD(CAST(`float64_col` AS BIGNUMERIC), CAST(0 AS BIGNUMERIC))
-  END AS `float_mod_0`
+  END AS `float_mod_0`,
+  NULL AS `float_mod_null`
 FROM `bigframes-dev`.`sqlglot_test`.`scalar_types` AS `bft_0`

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

@@ -241,5 +241,7 @@ SELECT
         ELSE 1
       END
     )
-  END AS `float_pow_1`
+  END AS `float_pow_1`,
+  NULL AS `float_pow_null`,
+  NULL AS `null_pow_float`
 FROM `bigframes-dev`.`sqlglot_test`.`scalar_types` AS `bft_0`

tests/unit/core/compile/sqlglot/expressions/snapshots/test_string_ops/test_str_slice/out.sql

@@ -13,6 +13,6 @@ SELECT
   SUBSTRING(
     `string_col`,
     GREATEST(1, LENGTH(`string_col`) + -2),
-    5 - GREATEST(0, LENGTH(`string_col`) + -3)
+    GREATEST(0, 5 - GREATEST(0, LENGTH(`string_col`) + -3))
   ) AS `m3_5`
 FROM `bigframes-dev`.`sqlglot_test`.`scalar_types` AS `bft_0`