Merge branch 'main' into more_hybrid_nodes

Author: TrevorBergeron

bigframes/core/compile/scalar_op_compiler.py

@@ -487,9 +487,9 @@ def isalpha_op_impl(x: ibis_types.Value):
 
 @scalar_op_compiler.register_unary_op(ops.isdigit_op)
 def isdigit_op_impl(x: ibis_types.Value):
-    # Based on docs, should include superscript/subscript-ed numbers
-    # Tests however pass only when set to Nd unicode class
-    return typing.cast(ibis_types.StringValue, x).re_search(r"^(\p{Nd})+$")
+    return typing.cast(ibis_types.StringValue, x).re_search(
+        r"^[\p{Nd}\x{00B9}\x{00B2}\x{00B3}\x{2070}\x{2074}-\x{2079}\x{2080}-\x{2089}]+$"
+    )
 
 
 @scalar_op_compiler.register_unary_op(ops.isdecimal_op)

tests/system/large/functions/test_remote_function.py

@@ -1707,7 +1707,7 @@ def analyze(row):
                 {
                     "dtype": row.dtype,
                     "count": row.count(),
-                    "min": row.max(),
+                    "min": row.min(),
                     "max": row.max(),
                     "mean": row.mean(),
                     "std": row.std(),

tests/system/small/operations/test_strings.py

@@ -324,13 +324,10 @@ def test_isalpha(weird_strings, weird_strings_pd):
     )
 
 
-@pytest.mark.skipif(
-    "dev" in pa.__version__,
-    # b/333484335 pyarrow is inconsistent on the behavior
-    reason="pyarrow dev version is inconsistent on isdigit behavior.",
-)
 def test_isdigit(weird_strings, weird_strings_pd):
-    pd_result = weird_strings_pd.str.isdigit()
+    # check the behavior against normal pandas str, since pyarrow has a bug with superscripts/fractions b/333484335
+    # astype object instead of str to support pd.NA
+    pd_result = weird_strings_pd.astype(object).str.isdigit()
     bf_result = weird_strings.str.isdigit().to_pandas()
 
     pd.testing.assert_series_equal(

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

@@ -1,16 +1,13 @@
 WITH `bfcte_0` AS (
   SELECT
-    `int64_col` AS `bfcol_0`,
-    `rowindex` AS `bfcol_1`
+    `int64_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
 ), `bfcte_1` AS (
   SELECT
     *,
-    `bfcol_1` AS `bfcol_4`,
-    `bfcol_0` + `bfcol_0` AS `bfcol_5`
+    `bfcol_0` + `bfcol_0` AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_4` AS `rowindex`,
-  `bfcol_5` AS `int64_col`
+  `bfcol_1` AS `int64_col`
 FROM `bfcte_1`

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

@@ -1,16 +1,13 @@
 WITH `bfcte_0` AS (
   SELECT
-    `int64_col` AS `bfcol_0`,
-    `rowindex` AS `bfcol_1`
+    `int64_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
 ), `bfcte_1` AS (
   SELECT
     *,
-    `bfcol_1` AS `bfcol_4`,
-    `bfcol_0` + 1 AS `bfcol_5`
+    `bfcol_0` + 1 AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_4` AS `rowindex`,
-  `bfcol_5` AS `int64_col`
+  `bfcol_1` AS `int64_col`
 FROM `bfcte_1`

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

@@ -1,16 +1,13 @@
 WITH `bfcte_0` AS (
   SELECT
-    `rowindex` AS `bfcol_0`,
-    `string_col` AS `bfcol_1`
+    `string_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`scalar_types`
 ), `bfcte_1` AS (
   SELECT
     *,
-    `bfcol_0` AS `bfcol_4`,
-    CONCAT(`bfcol_1`, 'a') AS `bfcol_5`
+    CONCAT(`bfcol_0`, 'a') AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_4` AS `rowindex`,
-  `bfcol_5` AS `string_col`
+  `bfcol_1` AS `string_col`
 FROM `bfcte_1`

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

@@ -1,20 +1,13 @@
 WITH `bfcte_0` AS (
   SELECT
-    `rowindex` AS `bfcol_0`,
-    `json_col` AS `bfcol_1`
+    `json_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`json_types`
 ), `bfcte_1` AS (
   SELECT
     *,
-    JSON_SET(`bfcol_1`, '$.a', 100) AS `bfcol_4`
+    JSON_SET(`bfcol_0`, '$.a', 100) AS `bfcol_1`
   FROM `bfcte_0`
-), `bfcte_2` AS (
-  SELECT
-    *,
-    JSON_SET(`bfcol_4`, '$.b', 'hi') AS `bfcol_7`
-  FROM `bfcte_1`
 )
 SELECT
-  `bfcol_0` AS `rowindex`,
-  `bfcol_7` AS `json_col`
-FROM `bfcte_2`
+  `bfcol_1` AS `json_col`
+FROM `bfcte_1`

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

@@ -1,15 +1,13 @@
 WITH `bfcte_0` AS (
   SELECT
-    `rowindex` AS `bfcol_0`,
-    `string_list_col` AS `bfcol_1`
+    `string_list_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`repeated_types`
 ), `bfcte_1` AS (
   SELECT
     *,
-    `bfcol_1`[SAFE_OFFSET(1)] AS `bfcol_4`
+    `bfcol_0`[SAFE_OFFSET(1)] AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_0` AS `rowindex`,
-  `bfcol_4` AS `string_list_col`
+  `bfcol_1` AS `string_list_col`
 FROM `bfcte_1`

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

@@ -1,21 +1,19 @@
 WITH `bfcte_0` AS (
   SELECT
-    `rowindex` AS `bfcol_0`,
-    `string_list_col` AS `bfcol_1`
+    `string_list_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`repeated_types`
 ), `bfcte_1` AS (
   SELECT
     *,
     ARRAY(
       SELECT
         el
-      FROM UNNEST(`bfcol_1`) AS el WITH OFFSET AS slice_idx
+      FROM UNNEST(`bfcol_0`) AS el WITH OFFSET AS slice_idx
       WHERE
         slice_idx >= 1
-    ) AS `bfcol_4`
+    ) AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_0` AS `rowindex`,
-  `bfcol_4` AS `string_list_col`
+  `bfcol_1` AS `string_list_col`
 FROM `bfcte_1`

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

@@ -1,21 +1,19 @@
 WITH `bfcte_0` AS (
   SELECT
-    `rowindex` AS `bfcol_0`,
-    `string_list_col` AS `bfcol_1`
+    `string_list_col` AS `bfcol_0`
   FROM `bigframes-dev`.`sqlglot_test`.`repeated_types`
 ), `bfcte_1` AS (
   SELECT
     *,
     ARRAY(
       SELECT
         el
-      FROM UNNEST(`bfcol_1`) AS el WITH OFFSET AS slice_idx
+      FROM UNNEST(`bfcol_0`) AS el WITH OFFSET AS slice_idx
       WHERE
         slice_idx >= 1 AND slice_idx < 5
-    ) AS `bfcol_4`
+    ) AS `bfcol_1`
   FROM `bfcte_0`
 )
 SELECT
-  `bfcol_0` AS `rowindex`,
-  `bfcol_4` AS `string_list_col`
+  `bfcol_1` AS `string_list_col`
 FROM `bfcte_1`