BigQuery: Modify client.dataset() to return DatasetReference instead of Dataset.

bigquery/google/cloud/bigquery/client.py

@@ -18,6 +18,7 @@
 from google.cloud.client import ClientWithProject
 from google.cloud.bigquery._http import Connection
 from google.cloud.bigquery.dataset import Dataset
+from google.cloud.bigquery.dataset import DatasetReference
 from google.cloud.bigquery.job import CopyJob
 from google.cloud.bigquery.job import ExtractJob
 from google.cloud.bigquery.job import LoadJob
@@ -148,7 +149,7 @@ def list_datasets(self, include_all=False, max_results=None,
             extra_params=extra_params)
 
     def dataset(self, dataset_name, project=None):
-        """Construct a dataset bound to this client.
+        """Construct a reference to a dataset.
 
         :type dataset_name: str
         :param dataset_name: Name of the dataset.
@@ -157,10 +158,13 @@ def dataset(self, dataset_name, project=None):
         :param project: (Optional) project ID for the dataset (defaults to
                         the project of the client).
 
-        :rtype: :class:`google.cloud.bigquery.dataset.Dataset`
-        :returns: a new ``Dataset`` instance
+        :rtype: :class:`google.cloud.bigquery.dataset.DatasetReference`
+        :returns: a new ``DatasetReference`` instance
         """
-        return Dataset(dataset_name, client=self, project=project)
+        if project is None:
+            project = self.project
+
+        return DatasetReference(project, dataset_name)
 
     def _get_query_results(self, job_id, project=None, timeout_ms=None):
         """Get the query results object for a query job.

bigquery/google/cloud/bigquery/dataset.py

@@ -19,6 +19,7 @@
 from google.cloud._helpers import _datetime_from_microseconds
 from google.cloud.exceptions import NotFound
 from google.cloud.bigquery.table import Table
+from google.cloud.bigquery.table import TableReference
 
 
 class AccessEntry(object):
@@ -126,6 +127,14 @@ def dataset_id(self):
         """
         return self._dataset_id
 
+    def table(self, table_id):
+        """Constructs a TableReference.
+
+        :rtype: :class:`google.cloud.bigquery.table.TableReference`
+        :returns: a TableReference for a table in this dataset.
+        """
+        return TableReference(self, table_id)
+
 
 class Dataset(object):
     """Datasets are containers for tables.

bigquery/google/cloud/bigquery/job.py

@@ -1362,8 +1362,8 @@ def _copy_configuration_properties(self, configuration):
             dest_local = self._destination_table_resource()
             if dest_remote != dest_local:
                 project = dest_remote['projectId']
-                dataset = self._client.dataset(
-                    dest_remote['datasetId'], project=project)
+                dataset = Dataset(
+                    dest_remote['datasetId'], self._client, project=project)
                 self.destination = dataset.table(dest_remote['tableId'])
 
         def_ds = configuration.get('defaultDataset')
@@ -1372,7 +1372,7 @@ def _copy_configuration_properties(self, configuration):
                 del self.default_dataset
         else:
             project = def_ds['projectId']
-            self.default_dataset = self._client.dataset(def_ds['datasetId'])
+            self.default_dataset = Dataset(def_ds['datasetId'], self._client)
 
         udf_resources = []
         for udf_mapping in configuration.get(self._UDF_KEY, ()):
@@ -1528,7 +1528,7 @@ def referenced_tables(self):
             ds_name = table['datasetId']
             t_dataset = datasets_by_project_name.get((t_project, ds_name))
             if t_dataset is None:
-                t_dataset = client.dataset(ds_name, project=t_project)
+                t_dataset = Dataset(ds_name, client, project=t_project)
                 datasets_by_project_name[(t_project, ds_name)] = t_dataset
 
             t_name = table['tableId']

bigquery/google/cloud/bigquery/table.py

@@ -48,6 +48,42 @@
 _DEFAULT_NUM_RETRIES = 6
 
 
+class TableReference(object):
+    """TableReferences are pointers to tables.
+
+    See
+    https://cloud.google.com/bigquery/docs/reference/rest/v2/tables
+
+    :type dataset_ref: :class:`google.cloud.bigquery.dataset.DatasetReference`
+    :param dataset_ref: a pointer to the dataset
+
+    :type table_id: str
+    :param table_id: the ID of the table
+    """
+
+    def __init__(self, dataset_ref, table_id):
+        self._dataset_ref = dataset_ref
+        self._table_id = table_id
+
+    @property
+    def dataset_ref(self):
+        """Pointer to the dataset.
+
+        :rtype: :class:`google.cloud.bigquery.dataset.DatasetReference`
+        :returns: a pointer to the dataset.
+        """
+        return self._dataset_ref
+
+    @property
+    def table_id(self):
+        """Table ID.
+
+        :rtype: str
+        :returns: the table ID.
+        """
+        return self._table_id
+
+
 class Table(object):
     """Tables represent a set of rows whose values correspond to a schema.
 

bigquery/tests/system.py

@@ -25,6 +25,7 @@
 import six
 
 from google.cloud import bigquery
+from google.cloud.bigquery.dataset import Dataset
 from google.cloud._helpers import UTC
 from google.cloud.bigquery import dbapi
 from google.cloud.exceptions import Forbidden
@@ -111,7 +112,7 @@ def _still_in_use(bad_request):
 
     def test_create_dataset(self):
         DATASET_NAME = _make_dataset_name('create_dataset')
-        dataset = Config.CLIENT.dataset(DATASET_NAME)
+        dataset = Dataset(DATASET_NAME, Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -122,20 +123,20 @@ def test_create_dataset(self):
 
     def test_reload_dataset(self):
         DATASET_NAME = _make_dataset_name('reload_dataset')
-        dataset = Config.CLIENT.dataset(DATASET_NAME)
+        dataset = Dataset(DATASET_NAME, Config.CLIENT)
         dataset.friendly_name = 'Friendly'
         dataset.description = 'Description'
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
 
-        other = Config.CLIENT.dataset(DATASET_NAME)
+        other = Dataset(DATASET_NAME, Config.CLIENT)
         other.reload()
         self.assertEqual(other.friendly_name, 'Friendly')
         self.assertEqual(other.description, 'Description')
 
     def test_patch_dataset(self):
-        dataset = Config.CLIENT.dataset(_make_dataset_name('patch_dataset'))
+        dataset = Dataset(_make_dataset_name('patch_dataset'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -149,7 +150,7 @@ def test_patch_dataset(self):
         self.assertEqual(dataset.description, 'Description')
 
     def test_update_dataset(self):
-        dataset = Config.CLIENT.dataset(_make_dataset_name('update_dataset'))
+        dataset = Dataset(_make_dataset_name('update_dataset'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -175,7 +176,7 @@ def test_list_datasets(self):
             'newest' + unique_resource_id(),
         ]
         for dataset_name in datasets_to_create:
-            created_dataset = Config.CLIENT.dataset(dataset_name)
+            created_dataset = Dataset(dataset_name, Config.CLIENT)
             retry_403(created_dataset.create)()
             self.to_delete.append(created_dataset)
 
@@ -189,7 +190,7 @@ def test_list_datasets(self):
         self.assertEqual(len(created), len(datasets_to_create))
 
     def test_create_table(self):
-        dataset = Config.CLIENT.dataset(_make_dataset_name('create_table'))
+        dataset = Dataset(_make_dataset_name('create_table'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -208,7 +209,7 @@ def test_create_table(self):
 
     def test_list_tables(self):
         DATASET_NAME = _make_dataset_name('list_tables')
-        dataset = Config.CLIENT.dataset(DATASET_NAME)
+        dataset = Dataset(DATASET_NAME, Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -244,7 +245,7 @@ def test_list_tables(self):
         self.assertEqual(len(created), len(tables_to_create))
 
     def test_patch_table(self):
-        dataset = Config.CLIENT.dataset(_make_dataset_name('patch_table'))
+        dataset = Dataset(_make_dataset_name('patch_table'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -266,7 +267,7 @@ def test_patch_table(self):
         self.assertEqual(table.description, 'Description')
 
     def test_update_table(self):
-        dataset = Config.CLIENT.dataset(_make_dataset_name('update_table'))
+        dataset = Dataset(_make_dataset_name('update_table'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -309,8 +310,8 @@ def test_insert_data_then_dump_table(self):
             ('Bhettye Rhubble', 27, None),
         ]
         ROW_IDS = range(len(ROWS))
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('insert_data_then_dump'))
+        dataset = Dataset(
+            _make_dataset_name('insert_data_then_dump'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()
@@ -353,8 +354,8 @@ def test_load_table_from_local_file_then_dump_table(self):
         ]
         TABLE_NAME = 'test_table'
 
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('load_local_then_dump'))
+        dataset = Dataset(
+            _make_dataset_name('load_local_then_dump'), Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -402,8 +403,8 @@ def test_load_table_from_local_avro_file_then_dump_table(self):
             ("orange", 590),
             ("red", 650)]
 
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('load_local_then_dump'))
+        dataset = Dataset(
+            _make_dataset_name('load_local_then_dump'), Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -466,8 +467,8 @@ def test_load_table_from_storage_then_dump_table(self):
 
         self.to_delete.insert(0, blob)
 
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('load_gcs_then_dump'))
+        dataset = Dataset(
+            _make_dataset_name('load_gcs_then_dump'), Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -536,8 +537,8 @@ def test_load_table_from_storage_w_autodetect_schema(self):
 
         self.to_delete.insert(0, blob)
 
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('load_gcs_then_dump'))
+        dataset = Dataset(
+            _make_dataset_name('load_gcs_then_dump'), Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -574,7 +575,7 @@ def test_job_cancel(self):
         TABLE_NAME = 'test_table'
         QUERY = 'SELECT * FROM %s.%s' % (DATASET_NAME, TABLE_NAME)
 
-        dataset = Config.CLIENT.dataset(DATASET_NAME)
+        dataset = Dataset(DATASET_NAME, Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -765,7 +766,7 @@ def test_dbapi_fetchall(self):
     def _load_table_for_dml(self, rows, dataset_name, table_name):
         from google.cloud._testing import _NamedTemporaryFile
 
-        dataset = Config.CLIENT.dataset(dataset_name)
+        dataset = Dataset(dataset_name, Config.CLIENT)
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
 
@@ -1081,7 +1082,7 @@ def test_dump_table_w_public_data(self):
         DATASET_NAME = 'samples'
         TABLE_NAME = 'natality'
 
-        dataset = Config.CLIENT.dataset(DATASET_NAME, project=PUBLIC)
+        dataset = Dataset(DATASET_NAME, Config.CLIENT, project=PUBLIC)
         table = dataset.table(TABLE_NAME)
         # Reload table to get the schema before fetching the rows.
         table.reload()
@@ -1134,8 +1135,8 @@ def test_insert_nested_nested(self):
             ('Some value', record)
         ]
         table_name = 'test_table'
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('issue_2951'))
+        dataset = Dataset(
+            _make_dataset_name('issue_2951'), Config.CLIENT)
 
         retry_403(dataset.create)()
         self.to_delete.append(dataset)
@@ -1154,8 +1155,8 @@ def test_insert_nested_nested(self):
     def test_create_table_insert_fetch_nested_schema(self):
 
         table_name = 'test_table'
-        dataset = Config.CLIENT.dataset(
-            _make_dataset_name('create_table_nested_schema'))
+        dataset = Dataset(
+            _make_dataset_name('create_table_nested_schema'), Config.CLIENT)
         self.assertFalse(dataset.exists())
 
         retry_403(dataset.create)()

bigquery/tests/unit/test_client.py

@@ -246,18 +246,31 @@ def test_list_datasets_explicit_response_missing_datasets_key(self):
         self.assertEqual(req['query_params'],
                          {'all': True, 'maxResults': 3, 'pageToken': TOKEN})
 
-    def test_dataset(self):
-        from google.cloud.bigquery.dataset import Dataset
+    def test_dataset_with_specified_project(self):
+        from google.cloud.bigquery.dataset import DatasetReference
+
+        PROJECT = 'PROJECT'
+        DATASET = 'dataset_name'
+        creds = _make_credentials()
+        http = object()
+        client = self._make_one(project=PROJECT, credentials=creds, _http=http)
+        dataset = client.dataset(DATASET, PROJECT)
+        self.assertIsInstance(dataset, DatasetReference)
+        self.assertEqual(dataset.dataset_id, DATASET)
+        self.assertEqual(dataset.project_id, PROJECT)
+
+    def test_dataset_with_default_project(self):
+        from google.cloud.bigquery.dataset import DatasetReference
 
         PROJECT = 'PROJECT'
         DATASET = 'dataset_name'
         creds = _make_credentials()
         http = object()
         client = self._make_one(project=PROJECT, credentials=creds, _http=http)
         dataset = client.dataset(DATASET)
-        self.assertIsInstance(dataset, Dataset)
-        self.assertEqual(dataset.name, DATASET)
-        self.assertIs(dataset._client, client)
+        self.assertIsInstance(dataset, DatasetReference)
+        self.assertEqual(dataset.dataset_id, DATASET)
+        self.assertEqual(dataset.project_id, PROJECT)
 
     def test_job_from_resource_unknown_type(self):
         PROJECT = 'PROJECT'

bigquery/tests/unit/test_dataset.py

@@ -101,6 +101,12 @@ def test_ctor_defaults(self):
         self.assertEqual(dataset_ref.project_id, 'some-project-1')
         self.assertEqual(dataset_ref.dataset_id, 'dataset_1')
 
+    def test_table(self):
+        dataset_ref = self._make_one('some-project-1', 'dataset_1')
+        table_ref = dataset_ref.table('table_1')
+        self.assertIs(table_ref.dataset_ref, dataset_ref)
+        self.assertEqual(table_ref.table_id, 'table_1')
+
 
 class TestDataset(unittest.TestCase):
     PROJECT = 'project'

bigquery/tests/unit/test_job.py

@@ -2697,11 +2697,6 @@ def __init__(self, project='project', connection=None):
         self.project = project
         self._connection = connection
 
-    def dataset(self, name, project=None):
-        from google.cloud.bigquery.dataset import Dataset
-
-        return Dataset(name, client=self, project=project)
-
     def _get_query_results(self, job_id):
         from google.cloud.bigquery.query import QueryResults