expose checksum as user kwarg

Author: Mikejmnez

ci/requirements/environment.yml

@@ -36,7 +36,7 @@ dependencies:
   - pooch
   - pre-commit
   - pyarrow # pandas raises a deprecation warning without this, breaking doctests
-  - pydap
+  # - pydap
   - pytest
   - pytest-asyncio
   - pytest-cov
@@ -65,3 +65,4 @@ dependencies:
       - jax # no way to get cpu-only jaxlib from conda if gpu is present
       - types-defusedxml
       - types-pexpect
+      - git+https://github.com/pydap/pydap.git # just for now - will restore to conda after new release

xarray/backends/pydap_.py

@@ -36,10 +36,11 @@
 
 
 class PydapArrayWrapper(BackendArray):
-    def __init__(self, array, batch=False, cache=None):
+    def __init__(self, array, batch=False, cache=None, checksums=True):
         self.array = array
         self._batch = batch
         self._cache = cache
+        self._checksums = checksums
 
     @property
     def shape(self) -> tuple[int, ...]:
@@ -63,7 +64,7 @@ def _getitem(self, key):
             from pydap.lib import resolve_batch_for_all_variables
 
             dataset = self.array.dataset
-            resolve_batch_for_all_variables(self.array, key)
+            resolve_batch_for_all_variables(self.array, key, checksums=self._checksums)
             result = np.asarray(
                 dataset._current_batch_promise.wait_for_result(self.array.id)
             )
@@ -98,7 +99,15 @@ class PydapDataStore(AbstractDataStore):
     be useful if the netCDF4 library is not available.
     """
 
-    def __init__(self, dataset, group=None, session=None, batch=False, protocol=None):
+    def __init__(
+        self,
+        dataset,
+        group=None,
+        session=None,
+        batch=False,
+        protocol=None,
+        checksums=True,
+    ):
         """
         Parameters
         ----------
@@ -113,6 +122,7 @@ def __init__(self, dataset, group=None, session=None, batch=False, protocol=None
         self._batch_done = False
         self._array_cache = {}  # holds 1D dimension data
         self._protocol = protocol
+        self._checksums = checksums  # true by default
 
     @classmethod
     def open(
@@ -126,6 +136,7 @@ def open(
         verify=None,
         user_charset=None,
         batch=False,
+        checksums=True,
     ):
         from pydap.client import open_url
         from pydap.net import DEFAULT_TIMEOUT
@@ -157,6 +168,7 @@ def open(
             # pydap dataset
             dataset = url.ds
         args = {"dataset": dataset}
+        args["checksums"] = checksums
         if group:
             args["group"] = group
         if url.startswith(("http", "dap2")):
@@ -202,7 +214,7 @@ def open_store_variable(self, var):
         else:
             # all non-dimension variables
             data = indexing.LazilyIndexedArray(
-                PydapArrayWrapper(var, self._batch, self._array_cache)
+                PydapArrayWrapper(var, self._batch, self._array_cache, self._checksums)
             )
 
         return Variable(dimensions, data, var.attributes)
@@ -256,7 +268,9 @@ def _get_data_array(self, var):
 
         if not self._batch_done or var.id not in self._array_cache:
             # store all dim data into a dict for reuse
-            self._array_cache = get_batch_data(var.parent, self._array_cache)
+            self._array_cache = get_batch_data(
+                var.parent, self._array_cache, self._checksums
+            )
             self._batch_done = True
 
         return self._array_cache[var.id]
@@ -305,6 +319,7 @@ def open_dataset(
         verify=None,
         user_charset=None,
         batch=False,
+        checksums=True,
     ) -> Dataset:
         store = PydapDataStore.open(
             url=filename_or_obj,
@@ -316,6 +331,7 @@ def open_dataset(
             verify=verify,
             user_charset=user_charset,
             batch=batch,
+            checksums=checksums,
         )
         store_entrypoint = StoreBackendEntrypoint()
         with close_on_error(store):
@@ -349,6 +365,7 @@ def open_datatree(
         verify=None,
         user_charset=None,
         batch=False,
+        checksums=True,
     ) -> DataTree:
         groups_dict = self.open_groups_as_dict(
             filename_or_obj,
@@ -366,6 +383,7 @@ def open_datatree(
             verify=application,
             user_charset=user_charset,
             batch=batch,
+            checksums=checksums,
         )
 
         return datatree_from_dict_with_io_cleanup(groups_dict)
@@ -388,6 +406,7 @@ def open_groups_as_dict(
         verify=None,
         user_charset=None,
         batch=False,
+        checksums=True,
     ) -> dict[str, Dataset]:
         from xarray.core.treenode import NodePath
 
@@ -400,6 +419,7 @@ def open_groups_as_dict(
             verify=verify,
             user_charset=user_charset,
             batch=batch,
+            checksums=checksums,
         )
 
         # Check for a group and make it a parent if it exists