Use DataLocator consistently on backend (#1208)

* allow DataLocator to accept another locator as init param

* migrate to DataLocator

* migrate to DataLocator

* lint

* migrate to DataLocator

* add check for erroroneous use of remote path and annotations

* lint

* revert default data location - now back go CWD

* remove unused import

server/data_common/data_adaptor.py

@@ -10,7 +10,6 @@ from server.common.errors import FilterError, JSONEncodingValueError
 from server.compute.diffexp import diffexp_ttest
 from server.common.utils import jsonify_numpy
 from server.common.app_config import AppFeature, AppConfig
-from server.common.data_locator import DataLocator
 
 
 class DataAdaptor(metaclass=ABCMeta):
@@ -32,17 +31,17 @@ class DataAdaptor(metaclass=ABCMeta):
 
     @staticmethod
     @abstractmethod
-    def pre_load_validation(location):
+    def pre_load_validation(data_locator):
         pass
 
     @staticmethod
     @abstractmethod
-    def open(location, config):
+    def open(data_locator, config):
         pass
 
     @staticmethod
     @abstractmethod
-    def file_size(location):
+    def file_size(data_locator):
         pass
 
     @abstractmethod
@@ -105,6 +104,10 @@ class DataAdaptor(metaclass=ABCMeta):
     def get_location(self):
         pass
 
+    @abstractmethod
+    def get_data_locator(self):
+        pass
+
     def get_about(self):
         return None
 
@@ -345,8 +348,7 @@ class DataAdaptor(metaclass=ABCMeta):
 
     def get_last_mod_time(self):
         try:
-            data_locator = DataLocator(self.get_location())
-            lastmod = data_locator.lastmodtime()
+            lastmod = self.get_data_locator().lastmodtime()
         except RuntimeError:
             lastmod = None
         return lastmod

server/data_common/matrix_loader.py

@@ -3,6 +3,7 @@ import threading
 import time
 from server.data_common.rwlock import RWLock
 from server.common.errors import DatasetAccessError
+from server.common.data_locator import DataLocator
 from contextlib import contextmanager
 
 
@@ -136,7 +137,8 @@ class MatrixDataType(Enum):
 
 class MatrixDataLoader(object):
     def __init__(self, location, etype=None):
-        self.location = location
+        """ location can be a string or DataLocator """
+        self.location = DataLocator(location)
         if etype is None:
             self.etype = self.matrix_data_type()
         else:
@@ -152,9 +154,9 @@ class MatrixDataLoader(object):
             self.matrix_type = CxgAdaptor
 
     def matrix_data_type(self):
-        if self.location.endswith(".h5ad"):
+        if self.location.path.endswith(".h5ad"):
             return MatrixDataType.H5AD
-        elif ".cxg" in self.location:
+        elif ".cxg" in self.location.path:
             return MatrixDataType.CXG
         else:
             return MatrixDataType.UNKNOWN