Merge remote-tracking branch 'origin/main' into refactor-isnull-op

Author: tswast

.pre-commit-config.yaml

@@ -46,4 +46,4 @@ repos:
     rev: v2.0.2
     hooks:
     -   id: biome-check
-        files: '\.js$'
+        files: '\.(js|css)$'

CHANGELOG.md

@@ -4,6 +4,52 @@
 
 [1]: https://pypi.org/project/bigframes/#history
 
+## [2.13.0](https://github.com/googleapis/python-bigquery-dataframes/compare/v2.12.0...v2.13.0) (2025-07-25)
+
+
+### Features
+
+* _read_gbq_colab creates hybrid session ([#1901](https://github.com/googleapis/python-bigquery-dataframes/issues/1901)) ([31b17b0](https://github.com/googleapis/python-bigquery-dataframes/commit/31b17b01706ccfcee9a2d838c43a9609ec4dc218))
+* Add CSS styling for TableWidget pagination interface ([#1934](https://github.com/googleapis/python-bigquery-dataframes/issues/1934)) ([5b232d7](https://github.com/googleapis/python-bigquery-dataframes/commit/5b232d7e33563196316f5dbb50b28c6be388d440))
+* Add row numbering local pushdown in hybrid execution ([#1932](https://github.com/googleapis/python-bigquery-dataframes/issues/1932)) ([92a2377](https://github.com/googleapis/python-bigquery-dataframes/commit/92a237712aa4ce516b1a44748127b34d7780fff6))
+* Implement Index.get_loc ([#1921](https://github.com/googleapis/python-bigquery-dataframes/issues/1921)) ([bbbcaf3](https://github.com/googleapis/python-bigquery-dataframes/commit/bbbcaf35df113617fd6bb8ae36468cf3f7ab493b))
+
+
+### Bug Fixes
+
+* Add license header and correct issues in dbt sample ([#1931](https://github.com/googleapis/python-bigquery-dataframes/issues/1931)) ([ab01b0a](https://github.com/googleapis/python-bigquery-dataframes/commit/ab01b0a236ffc7b667f258e0497105ea5c3d3aab))
+
+
+### Dependencies
+
+* Replace `google-cloud-iam` with `grpc-google-iam-v1` ([#1864](https://github.com/googleapis/python-bigquery-dataframes/issues/1864)) ([e5ff8f7](https://github.com/googleapis/python-bigquery-dataframes/commit/e5ff8f7d9fdac3ea47dabcc80a2598d601f39e64))
+
+## [2.12.0](https://github.com/googleapis/python-bigquery-dataframes/compare/v2.11.0...v2.12.0) (2025-07-23)
+
+
+### Features
+
+* Add code samples for dbt bigframes integration ([#1898](https://github.com/googleapis/python-bigquery-dataframes/issues/1898)) ([7e03252](https://github.com/googleapis/python-bigquery-dataframes/commit/7e03252d31e505731db113eb38af77842bf29b9b))
+* Add isin local execution to hybrid engine ([#1915](https://github.com/googleapis/python-bigquery-dataframes/issues/1915)) ([c0cefd3](https://github.com/googleapis/python-bigquery-dataframes/commit/c0cefd36cfd55962b86178d2a612d625ed17f79c))
+* Add ml.metrics.mean_absolute_error method ([#1910](https://github.com/googleapis/python-bigquery-dataframes/issues/1910)) ([15b8449](https://github.com/googleapis/python-bigquery-dataframes/commit/15b8449dc5ad0c8190a5cbf47894436de18c8e88))
+* Allow local arithmetic execution in hybrid engine ([#1906](https://github.com/googleapis/python-bigquery-dataframes/issues/1906)) ([ebdcd02](https://github.com/googleapis/python-bigquery-dataframes/commit/ebdcd0240f0d8edaef3094b3a4e664b4a84d4a25))
+* Provide day_of_year and day_of_week for dt accessor ([#1911](https://github.com/googleapis/python-bigquery-dataframes/issues/1911)) ([40e7638](https://github.com/googleapis/python-bigquery-dataframes/commit/40e76383948a79bde48108f6180fd6ae2b3d0875))
+* Support params `max_batching_rows`, `container_cpu`, and `container_memory` for `udf` ([#1897](https://github.com/googleapis/python-bigquery-dataframes/issues/1897)) ([8baa912](https://github.com/googleapis/python-bigquery-dataframes/commit/8baa9126e595ae682469a6bb462244240699f57f))
+* Support typed pyarrow.Scalar in assignment  ([#1930](https://github.com/googleapis/python-bigquery-dataframes/issues/1930)) ([cd28e12](https://github.com/googleapis/python-bigquery-dataframes/commit/cd28e12b3f70a6934a68963a7f25dbd5e3c67335))
+
+
+### Bug Fixes
+
+* Correct min field from max() to min() in remote function tests ([#1917](https://github.com/googleapis/python-bigquery-dataframes/issues/1917)) ([d5c54fc](https://github.com/googleapis/python-bigquery-dataframes/commit/d5c54fca32ed75c1aef52c99781db7f8ac7426e1))
+* Resolve location reset issue in bigquery options ([#1914](https://github.com/googleapis/python-bigquery-dataframes/issues/1914)) ([c15cb8a](https://github.com/googleapis/python-bigquery-dataframes/commit/c15cb8a1a9c834c2c1c2984930415b246f3f948b))
+* Series.str.isdigit in unicode superscripts and fractions ([#1924](https://github.com/googleapis/python-bigquery-dataframes/issues/1924)) ([8d46c36](https://github.com/googleapis/python-bigquery-dataframes/commit/8d46c36da7881a99861166c03a0831beff8ee0dd))
+
+
+### Documentation
+
+* Add code snippets for session and IO public docs ([#1919](https://github.com/googleapis/python-bigquery-dataframes/issues/1919)) ([6e01cbe](https://github.com/googleapis/python-bigquery-dataframes/commit/6e01cbec0dcf40e528b4a96e944681df18773c11))
+* Add snippets for performance optimization doc ([#1923](https://github.com/googleapis/python-bigquery-dataframes/issues/1923)) ([4da309e](https://github.com/googleapis/python-bigquery-dataframes/commit/4da309e27bd58a685e8aca953717da75d4ba5305))
+
 ## [2.11.0](https://github.com/googleapis/python-bigquery-dataframes/compare/v2.10.0...v2.11.0) (2025-07-15)
 
 

MANIFEST.in

@@ -17,7 +17,7 @@
 # Generated by synthtool. DO NOT EDIT!
 include README.rst LICENSE
 recursive-include third_party/bigframes_vendored *
-recursive-include bigframes *.json *.proto *.js py.typed
+recursive-include bigframes *.json *.proto *.js *.css py.typed
 recursive-include tests *
 global-exclude *.py[co]
 global-exclude __pycache__

bigframes/_config/bigquery_options.py

@@ -171,7 +171,7 @@ def location(self) -> Optional[str]:
 
     @location.setter
     def location(self, value: Optional[str]):
-        if self._session_started and self._location != value:
+        if self._session_started and self._location != _get_validated_location(value):
             raise ValueError(SESSION_STARTED_MESSAGE.format(attribute="location"))
         self._location = _get_validated_location(value)
 

bigframes/core/compile/ibis_compiler/scalar_op_registry.py

@@ -300,9 +300,9 @@ 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)
@@ -1311,7 +1311,7 @@ def eq_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x == y
 
 
@@ -1321,7 +1321,7 @@ def eq_nulls_match_op(
     y: ibis_types.Value,
 ):
     """Variant of eq_op where nulls match each other. Only use where dtypes are known to be same."""
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     literal = ibis_types.literal("$NULL_SENTINEL$")
     if hasattr(x, "fill_null"):
         left = x.cast(ibis_dtypes.str).fill_null(literal)
@@ -1338,7 +1338,7 @@ def ne_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x != y
 
 
@@ -1350,13 +1350,10 @@ def _null_or_value(value: ibis_types.Value, where_value: ibis_types.BooleanValue
     )
 
 
-def _coerce_comparables(
-    x: ibis_types.Value,
-    y: ibis_types.Value,
-):
-    if x.type().is_boolean() and not y.type().is_boolean():
+def _coerce_bools(x: ibis_types.Value, y: ibis_types.Value, *, always: bool = False):
+    if x.type().is_boolean() and (always or not y.type().is_boolean()):
         x = x.cast(ibis_dtypes.int64)
-    elif y.type().is_boolean() and not x.type().is_boolean():
+    if y.type().is_boolean() and (always or not x.type().is_boolean()):
         y = y.cast(ibis_dtypes.int64)
     return x, y
 
@@ -1417,8 +1414,18 @@ def add_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    x, y = _coerce_bools(x, y)
     if isinstance(x, ibis_types.NullScalar) or isinstance(x, ibis_types.NullScalar):
         return ibis_types.null()
+
+    if x.type().is_boolean() and y.type().is_boolean():
+        x, y = _coerce_bools(x, y, always=True)
+        return (
+            typing.cast(ibis_types.NumericValue, x)
+            + typing.cast(ibis_types.NumericValue, x)
+        ).cast(ibis_dtypes.Boolean)
+
+    x, y = _coerce_bools(x, y)
     return x + y  # type: ignore
 
 
@@ -1428,6 +1435,7 @@ def sub_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    x, y = _coerce_bools(x, y)
     return typing.cast(ibis_types.NumericValue, x) - typing.cast(
         ibis_types.NumericValue, y
     )
@@ -1439,6 +1447,13 @@ def mul_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    if x.type().is_boolean() and y.type().is_boolean():
+        x, y = _coerce_bools(x, y, always=True)
+        return (
+            typing.cast(ibis_types.NumericValue, x)
+            * typing.cast(ibis_types.NumericValue, x)
+        ).cast(ibis_dtypes.Boolean)
+    x, y = _coerce_bools(x, y)
     return typing.cast(ibis_types.NumericValue, x) * typing.cast(
         ibis_types.NumericValue, y
     )
@@ -1450,6 +1465,7 @@ def div_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    x, y = _coerce_bools(x, y)
     return typing.cast(ibis_types.NumericValue, x) / typing.cast(
         ibis_types.NumericValue, y
     )
@@ -1461,6 +1477,7 @@ def pow_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    x, y = _coerce_bools(x, y)
     if x.type().is_integer() and y.type().is_integer():
         return _int_pow_op(x, y)
     else:
@@ -1474,6 +1491,7 @@ def unsafe_pow_op(
     y: ibis_types.Value,
 ):
     """For internal use only - where domain and overflow checks are not needed."""
+    x, y = _coerce_bools(x, y)
     return typing.cast(ibis_types.NumericValue, x) ** typing.cast(
         ibis_types.NumericValue, y
     )
@@ -1562,7 +1580,7 @@ def lt_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x < y
 
 
@@ -1572,7 +1590,7 @@ def le_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x <= y
 
 
@@ -1582,7 +1600,7 @@ def gt_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x > y
 
 
@@ -1592,7 +1610,7 @@ def ge_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
-    x, y = _coerce_comparables(x, y)
+    x, y = _coerce_bools(x, y)
     return x >= y
 
 
@@ -1602,6 +1620,10 @@ def floordiv_op(
     x: ibis_types.Value,
     y: ibis_types.Value,
 ):
+    if x.type().is_boolean():
+        x = x.cast(ibis_dtypes.int64)
+    elif y.type().is_boolean():
+        y = y.cast(ibis_dtypes.int64)
     x_numeric = typing.cast(ibis_types.NumericValue, x)
     y_numeric = typing.cast(ibis_types.NumericValue, y)
     floordiv_expr = x_numeric // y_numeric
@@ -1640,6 +1662,7 @@ def mod_op(
     if isinstance(op, ibis_generic.Literal) and op.value == 0:
         return ibis_types.null().cast(x.type())
 
+    x, y = _coerce_bools(x, y)
     if x.type().is_integer() and y.type().is_integer():
         # both are ints, no casting necessary
         return _int_mod(

bigframes/core/compile/polars/compiler.py

@@ -35,6 +35,7 @@
 import bigframes.operations.comparison_ops as comp_ops
 import bigframes.operations.generic_ops as gen_ops
 import bigframes.operations.numeric_ops as num_ops
+import bigframes.operations.string_ops as string_ops
 
 polars_installed = True
 if TYPE_CHECKING:
@@ -146,6 +147,14 @@ def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
         def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
             return input.abs()
 
+        @compile_op.register(num_ops.FloorOp)
+        def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
+            return input.floor()
+
+        @compile_op.register(num_ops.CeilOp)
+        def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
+            return input.ceil()
+
         @compile_op.register(num_ops.PosOp)
         def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
             return input.__pos__()
@@ -182,10 +191,6 @@ def _(self, op: ops.ScalarOp, l_input: pl.Expr, r_input: pl.Expr) -> pl.Expr:
         def _(self, op: ops.ScalarOp, l_input: pl.Expr, r_input: pl.Expr) -> pl.Expr:
             return l_input // r_input
 
-        @compile_op.register(num_ops.FloorDivOp)
-        def _(self, op: ops.ScalarOp, l_input: pl.Expr, r_input: pl.Expr) -> pl.Expr:
-            return l_input // r_input
-
         @compile_op.register(num_ops.ModOp)
         def _(self, op: ops.ScalarOp, l_input: pl.Expr, r_input: pl.Expr) -> pl.Expr:
             return l_input % r_input
@@ -262,6 +267,11 @@ def _(self, op: ops.ScalarOp, input: pl.Expr) -> pl.Expr:
             # eg. We want "True" instead of "true" for bool to strin
             return input.cast(_DTYPE_MAPPING[op.to_type], strict=not op.safe)
 
+        @compile_op.register(string_ops.StrConcatOp)
+        def _(self, op: ops.ScalarOp, l_input: pl.Expr, r_input: pl.Expr) -> pl.Expr:
+            assert isinstance(op, string_ops.StrConcatOp)
+            return pl.concat_str(l_input, r_input)
+
     @dataclasses.dataclass(frozen=True)
     class PolarsAggregateCompiler:
         scalar_compiler = PolarsExpressionCompiler()
@@ -495,6 +505,30 @@ def compile_join(self, node: nodes.JoinNode):
             left, right, node.type, left_on, right_on, node.joins_nulls
         )
 
+    @compile_node.register
+    def compile_isin(self, node: nodes.InNode):
+        left = self.compile_node(node.left_child)
+        right = self.compile_node(node.right_child).unique(node.right_col.id.sql)
+        right = right.with_columns(pl.lit(True).alias(node.indicator_col.sql))
+
+        left_ex, right_ex = lowering._coerce_comparables(node.left_col, node.right_col)
+
+        left_pl_ex = self.expr_compiler.compile_expression(left_ex)
+        right_pl_ex = self.expr_compiler.compile_expression(right_ex)
+
+        joined = left.join(
+            right,
+            how="left",
+            left_on=left_pl_ex,
+            right_on=right_pl_ex,
+            # Note: join_nulls renamed to nulls_equal for polars 1.24
+            join_nulls=node.joins_nulls,  # type: ignore
+            coalesce=False,
+        )
+        passthrough = [pl.col(id) for id in left.columns]
+        indicator = pl.col(node.indicator_col.sql).fill_null(False)
+        return joined.select((*passthrough, indicator))
+
     def _ordered_join(
         self,
         left_frame: pl.LazyFrame,