When reading annotations from tiledb, check if the values are byte literals and if so, decode them. Also pin s3f3 to 0.4.2. (#1788)

server/common/annotations/hosted_tiledb.py

@@ -74,6 +74,10 @@ class AnnotationsHostedTileDB(Annotations):
         indexes = list()
 
         for col_name, col_val in data.items():
+            # If the column values are byte literals, decode them
+            if isinstance(col_val[0], bytes):
+                col_val = [value.decode('utf-8') for value in col_val]
+
             if repr_meta and col_name in repr_meta:
                 new_col = pd.Series(col_val, dtype=repr_meta[col_name])
                 data[col_name] = new_col

server/requirements.txt

@@ -20,5 +20,5 @@ scipy>=1.3.0
 requests>=2.22.0
 sqlalchemy>=1.3.18
 tiledb>=0.5.9,>=0.6.2
-s3fs>=0.4.2
+s3fs==0.4.2
 gunicorn>=20.0.4

server/test/unit/common/test_writable_annotation.py

@@ -1,22 +1,20 @@
 import json
-from os import path, listdir
+import shutil
 import unittest
+from os import path, listdir
 from unittest.mock import MagicMock, patch
 
+import numpy as np
+import pandas as pd
 import tiledb
 from flask import Flask
 
 import server.test.unit.decode_fbs as decode_fbs
-import shutil
+from server.common.errors import AnnotationCategoryNameError
-
-import numpy as np
-import pandas as pd
-
 from server.common.rest import schema_get_helper, annotations_put_fbs_helper
+from server.data_common.matrix_loader import MatrixDataType
 from server.db.cellxgene_orm import CellxGeneDataset, Annotation
 from server.test import data_with_tmp_annotations, make_fbs, data_with_tmp_tiledb_annotations
-from server.data_common.matrix_loader import MatrixDataType
-from server.common.errors import AnnotationCategoryNameError
 
 
 class auth(object):
@@ -77,7 +75,6 @@ class WritableTileDBStoredAnnotationTest(unittest.TestCase):
 
     def test_write_labels_creates_a_dataset_if_it_doesnt_exist(self):
         with self.app.test_request_context():
-
             new_name = 'new_dataset/location'
             self.data.get_location = MagicMock(return_value=new_name)
             num_datasets = len(self.db.query([CellxGeneDataset]))