Add 'QueryJob.referenced_tables' property. (#3801)

Author: tseaver

bigquery/google/cloud/bigquery/job.py

@@ -1387,6 +1387,36 @@ def statement_type(self):
         """
         return self._job_statistics().get('statementType')
 
+    @property
+    def referenced_tables(self):
+        """Return referenced tables from job statistics, if present.
+
+        See:
+        https://cloud.google.com/bigquery/docs/reference/rest/v2/jobs#statistics.query.referencedTables
+
+        :rtype: list of dict
+        :returns: mappings describing the query plan, or an empty list
+                  if the query has not yet completed.
+        """
+        tables = []
+        client = self._require_client(None)
+        datasets_by_project_name = {}
+
+        for table in self._job_statistics().get('referencedTables', ()):
+
+            t_project = table['projectId']
+
+            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)
+                datasets_by_project_name[(t_project, ds_name)] = t_dataset
+
+            t_name = table['tableId']
+            tables.append(t_dataset.table(t_name))
+
+        return tables
+
     def query_results(self):
         """Construct a QueryResults instance, bound to this job.
 

bigquery/tests/unit/test_job.py

@@ -1833,6 +1833,60 @@ def test_statement_type(self):
         query_stats['statementType'] = statement_type
         self.assertEqual(job.statement_type, statement_type)
 
+    def test_referenced_tables(self):
+        from google.cloud.bigquery.dataset import Dataset
+        from google.cloud.bigquery.table import Table
+
+        ref_tables_resource = [{
+            'projectId': self.PROJECT,
+            'datasetId': 'dataset',
+            'tableId': 'local1',
+        }, {
+
+            'projectId': self.PROJECT,
+            'datasetId': 'dataset',
+            'tableId': 'local2',
+        }, {
+
+            'projectId': 'other-project-123',
+            'datasetId': 'other-dataset',
+            'tableId': 'other-table',
+        }]
+        client = _Client(self.PROJECT)
+        job = self._make_one(self.JOB_NAME, self.QUERY, client)
+        self.assertEqual(job.referenced_tables, [])
+
+        statistics = job._properties['statistics'] = {}
+        self.assertEqual(job.referenced_tables, [])
+
+        query_stats = statistics['query'] = {}
+        self.assertEqual(job.referenced_tables, [])
+
+        query_stats['referencedTables'] = ref_tables_resource
+
+        local1, local2, remote = job.referenced_tables
+
+        self.assertIsInstance(local1, Table)
+        self.assertEqual(local1.name, 'local1')
+        self.assertIsInstance(local1._dataset, Dataset)
+        self.assertEqual(local1.dataset_name, 'dataset')
+        self.assertEqual(local1.project, self.PROJECT)
+        self.assertIs(local1._dataset._client, client)
+
+        self.assertIsInstance(local2, Table)
+        self.assertEqual(local2.name, 'local2')
+        self.assertIsInstance(local2._dataset, Dataset)
+        self.assertEqual(local2.dataset_name, 'dataset')
+        self.assertEqual(local2.project, self.PROJECT)
+        self.assertIs(local2._dataset._client, client)
+
+        self.assertIsInstance(remote, Table)
+        self.assertEqual(remote.name, 'other-table')
+        self.assertIsInstance(remote._dataset, Dataset)
+        self.assertEqual(remote.dataset_name, 'other-dataset')
+        self.assertEqual(remote.project, 'other-project-123')
+        self.assertIs(remote._dataset._client, client)
+
     def test_query_results(self):
         from google.cloud.bigquery.query import QueryResults
 
@@ -2490,10 +2544,10 @@ def __init__(self, project='project', connection=None):
         self.project = project
         self._connection = connection
 
-    def dataset(self, name):
+    def dataset(self, name, project=None):
         from google.cloud.bigquery.dataset import Dataset
 
-        return Dataset(name, client=self)
+        return Dataset(name, client=self, project=project)
 
     def _get_query_results(self, job_id):
         from google.cloud.bigquery.query import QueryResults