diff --git a/automl/google/cloud/automl_v1beta1/tables/gcs_client.py b/automl/google/cloud/automl_v1beta1/tables/gcs_client.py index e5de17c3b0c9..99d40da2867c 100644 --- a/automl/google/cloud/automl_v1beta1/tables/gcs_client.py +++ b/automl/google/cloud/automl_v1beta1/tables/gcs_client.py @@ -132,7 +132,12 @@ def upload_pandas_dataframe(self, dataframe, uploaded_csv_name=None): uploaded_csv_name = "automl-tables-dataframe-{}.csv".format( int(time.time()) ) - csv_string = dataframe.to_csv() + + # Setting index to False to ignore exporting the data index: + # 1. The resulting column name for the index column is empty, AutoML + # Tables does not allow empty column name + # 2. The index is not an useful training information + csv_string = dataframe.to_csv(index=False) bucket = self.client.get_bucket(self.bucket_name) blob = bucket.blob(uploaded_csv_name) diff --git a/automl/tests/unit/gapic/v1beta1/test_gcs_client_v1beta1.py b/automl/tests/unit/gapic/v1beta1/test_gcs_client_v1beta1.py index f7a2e27ab7d8..222fca3244ee 100644 --- a/automl/tests/unit/gapic/v1beta1/test_gcs_client_v1beta1.py +++ b/automl/tests/unit/gapic/v1beta1/test_gcs_client_v1beta1.py @@ -139,7 +139,7 @@ def test_upload_pandas_dataframe(self): gcs_client.client.get_bucket.assert_called_with("my-bucket") mock_bucket.blob.assert_called_with("my-file.csv") - mock_blob.upload_from_string.assert_called_with(",col1,col2\n0,1,3\n1,2,4\n") + mock_blob.upload_from_string.assert_called_with("col1,col2\n1,3\n2,4\n") assert gcs_uri == "gs://my-bucket/my-file.csv" def test_upload_pandas_dataframe_no_csv_name(self): @@ -156,7 +156,7 @@ def test_upload_pandas_dataframe_no_csv_name(self): gcs_client.client.get_bucket.assert_called_with("my-bucket") mock_bucket.blob.assert_called_with(generated_csv_name) - mock_blob.upload_from_string.assert_called_with(",col1,col2\n0,1,3\n1,2,4\n") + mock_blob.upload_from_string.assert_called_with("col1,col2\n1,3\n2,4\n") assert re.match("^gs://my-bucket/automl-tables-dataframe-[0-9]*.csv$", gcs_uri) def test_upload_pandas_dataframe_not_type_dataframe(self):