Merge pull request pandas-dev#760 from shashank88/env_var_pandas

Allow setting force_bytes_to_unicode from env

Author: shashank88

arctic/_config.py

@@ -100,3 +100,8 @@ class FwPointersCfg(Enum):
 ENABLE_CACHE = not bool(os.environ.get('DISABLE_CACHE'))
 CACHE_COLL = 'cache'
 CACHE_DB = 'meta_db'
+
+# -------------------------------
+# Flag used to convert byte column/index/column names to unicode when read back.
+# -------------------------------
+FORCE_BYTES_TO_UNICODE = bool(os.environ.get('FORCE_BYTES_TO_UNICODE'))

arctic/store/_pandas_ndarray_store.py

@@ -9,6 +9,7 @@
 from arctic.serialization.numpy_records import SeriesSerializer, DataFrameSerializer
 from ._ndarray_store import NdarrayStore
 from .._compression import compress, decompress
+from .._config import FORCE_BYTES_TO_UNICODE
 from ..date._util import to_pandas_closed_closed
 from ..exceptions import ArcticException
 
@@ -197,7 +198,9 @@ def append(self, arctic_lib, version, symbol, item, previous_version, **kwargs):
 
     def read(self, arctic_lib, version, symbol, **kwargs):
         item = super(PandasDataFrameStore, self).read(arctic_lib, version, symbol, **kwargs)
-        return self.SERIALIZER.deserialize(item, force_bytes_to_unicode=kwargs.get('force_bytes_to_unicode', False))
+        # Try to check if force_bytes_to_unicode is set in kwargs else use the config value (which defaults to False)
+        force_bytes_to_unicode = kwargs.get('force_bytes_to_unicode', FORCE_BYTES_TO_UNICODE)
+        return self.SERIALIZER.deserialize(item, force_bytes_to_unicode=force_bytes_to_unicode)
 
     def read_options(self):
         return super(PandasDataFrameStore, self).read_options()