fix: generate GoogleSQL instead of legacy SQL data types for dry_run=True from bpd._read_gbq_colab with local pandas DataFrame (#1867)

* fix: generate GoogleSQL instead of legacy SQL data types for `dry_run=True` from `bpd._read_gbq_colab` with local pandas DataFrame

* finish adding system tests

* add unit tests

* map legacy sql types to googlesql

* fix more unit tests

* fix mypy

* fix python 3.9 tests shapely

Author: tswast

bigframes/core/tools/bigquery_schema.py

@@ -18,6 +18,12 @@
 
 import google.cloud.bigquery
 
+_LEGACY_TO_GOOGLESQL_TYPES = {
+    "BOOLEAN": "BOOL",
+    "INTEGER": "INT64",
+    "FLOAT": "FLOAT64",
+}
+
 
 def _type_to_sql(field: google.cloud.bigquery.SchemaField):
     """Turn the type information of the field into SQL.
@@ -26,7 +32,12 @@ def _type_to_sql(field: google.cloud.bigquery.SchemaField):
     """
     if field.field_type.casefold() in ("record", "struct"):
         return _to_struct(field.fields)
-    return field.field_type
+
+    # Map from legacy SQL names (the ones used in the BigQuery schema API) to
+    # the GoogleSQL types. Importantly, FLOAT is from legacy SQL, but not valid
+    # in GoogleSQL. See internal issue b/428190014.
+    type_ = _LEGACY_TO_GOOGLESQL_TYPES.get(field.field_type.upper(), field.field_type)
+    return type_
 
 
 def _field_to_sql(field: google.cloud.bigquery.SchemaField):

bigframes/pandas/io/api.py

@@ -218,8 +218,27 @@ def read_gbq(
 read_gbq.__doc__ = inspect.getdoc(bigframes.session.Session.read_gbq)
 
 
+def _run_read_gbq_colab_sessionless_dry_run(
+    query: str,
+    *,
+    pyformat_args: Dict[str, Any],
+) -> pandas.Series:
+    """Run a dry_run without a session."""
+
+    query_formatted = bigframes.core.pyformat.pyformat(
+        query,
+        pyformat_args=pyformat_args,
+        dry_run=True,
+    )
+    bqclient = _get_bqclient()
+    job = _dry_run(query_formatted, bqclient)
+    return dry_runs.get_query_stats_with_inferred_dtypes(job, (), ())
+
+
 def _try_read_gbq_colab_sessionless_dry_run(
-    create_query: Callable[[], str],
+    query: str,
+    *,
+    pyformat_args: Dict[str, Any],
 ) -> Optional[pandas.Series]:
     """Run a dry_run without a session, only if the session hasn't yet started."""
 
@@ -230,10 +249,9 @@ def _try_read_gbq_colab_sessionless_dry_run(
     # to local data and not any BigQuery tables.
     with _default_location_lock:
         if not config.options.bigquery._session_started:
-            bqclient = _get_bqclient()
-            query = create_query()
-            job = _dry_run(query, bqclient)
-            return dry_runs.get_query_stats_with_inferred_dtypes(job, (), ())
+            return _run_read_gbq_colab_sessionless_dry_run(
+                query, pyformat_args=pyformat_args
+            )
 
     # Explicitly return None to indicate that we didn't run the dry run query.
     return None
@@ -286,21 +304,13 @@ def _read_gbq_colab(
     if pyformat_args is None:
         pyformat_args = {}
 
-    # Delay formatting the query with the special "session-less" logic. This
-    # avoids doing unnecessary work if the session already has a location or has
-    # already started.
-    create_query = functools.partial(
-        bigframes.core.pyformat.pyformat,
-        query_or_table,
-        pyformat_args=pyformat_args,
-        dry_run=True,
-    )
-
     # Only try to set the global location if it's not a dry run. We don't want
     # to bind to a location too early. This is especially important if the query
     # only refers to local data and not any BigQuery tables.
     if dry_run:
-        result = _try_read_gbq_colab_sessionless_dry_run(create_query)
+        result = _try_read_gbq_colab_sessionless_dry_run(
+            query_or_table, pyformat_args=pyformat_args
+        )
 
         if result is not None:
             return result
@@ -309,6 +319,15 @@ def _read_gbq_colab(
         # started. That means we can safely call the "real" _read_gbq_colab,
         # which generates slightly nicer SQL.
     else:
+        # Delay formatting the query with the special "session-less" logic. This
+        # avoids doing unnecessary work if the session already has a location or has
+        # already started.
+        create_query = functools.partial(
+            bigframes.core.pyformat.pyformat,
+            query_or_table,
+            pyformat_args=pyformat_args,
+            dry_run=True,
+        )
         _set_default_session_location_if_possible_deferred_query(create_query)
 
     return global_session.with_default_session(

tests/system/small/pandas/io/__init__.py

@@ -0,0 +1,13 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.

tests/system/small/pandas/io/api/__init__.py

@@ -0,0 +1,13 @@
+# Copyright 2025 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.