Merge pull request #95 from DataONEorg/feature-94-client-bugfix

Feature-94: Client Bug Fixes

src/hashstore/filehashstore.py

@@ -105,6 +105,8 @@ def __init__(self, properties=None):
             self.objects = self.root + "/objects"
             self.metadata = self.root + "/metadata"
             self.refs = self.root + "/refs"
+            self.cids = self.refs + "/cids"
+            self.pids = self.refs + "/pids"
             if not os.path.exists(self.objects):
                 self._create_path(self.objects + "/tmp")
             if not os.path.exists(self.metadata):
@@ -128,7 +130,7 @@ def __init__(self, properties=None):
     # Configuration and Related Methods
 
     @staticmethod
-    def _load_properties(hahstore_yaml_path, hashstore_required_prop_keys):
+    def _load_properties(hashstore_yaml_path, hashstore_required_prop_keys):
         """Get and return the contents of the current HashStore configuration.
 
         :return: HashStore properties with the following keys (and values):
@@ -138,7 +140,7 @@ def _load_properties(hahstore_yaml_path, hashstore_required_prop_keys):
             - ``store_metadata_namespace`` (str): Namespace for the HashStore's system metadata.
         :rtype: dict
         """
-        if not os.path.exists(hahstore_yaml_path):
+        if not os.path.exists(hashstore_yaml_path):
             exception_string = (
                 "FileHashStore - load_properties: hashstore.yaml not found"
                 + " in store root path."
@@ -147,7 +149,7 @@ def _load_properties(hahstore_yaml_path, hashstore_required_prop_keys):
             raise FileNotFoundError(exception_string)
 
         # Open file
-        with open(hahstore_yaml_path, "r", encoding="utf-8") as hs_yaml_file:
+        with open(hashstore_yaml_path, "r", encoding="utf-8") as hs_yaml_file:
             yaml_data = yaml.safe_load(hs_yaml_file)
 
         # Get hashstore properties
@@ -2167,9 +2169,9 @@ def _get_store_path(self, entity):
         elif entity == "refs":
             return Path(self.refs)
         elif entity == "cid":
-            return Path(self.refs) / "cid"
+            return Path(self.cids)
         elif entity == "pid":
-            return Path(self.refs) / "pid"
+            return Path(self.pids)
         else:
             raise ValueError(
                 f"entity: {entity} does not exist. Do you mean 'objects', 'metadata' or 'refs'?"
@@ -2210,9 +2212,9 @@ def _count(self, entity):
         elif entity == "metadata":
             directory_to_count = self.metadata
         elif entity == "pid":
-            directory_to_count = self.refs + "/pid"
+            directory_to_count = self.pids
         elif entity == "cid":
-            directory_to_count = self.refs + "/cid"
+            directory_to_count = self.cids
         else:
             raise ValueError(
                 f"entity: {entity} does not exist. Do you mean 'objects' or 'metadata'?"

src/hashstore/hashstoreclient.py

@@ -1,4 +1,5 @@
 """HashStore Command Line App"""
+
 import logging
 import os
 from argparse import ArgumentParser
@@ -741,6 +742,13 @@ def main():
             f"Missing config file (hashstore.yaml) at store path: {store_path}."
             + " HashStore must first be initialized, use `--help` for more information."
         )
+    else:
+        # Get the default format_id for sysmeta
+        with open(store_path_config_yaml, "r", encoding="utf-8") as hs_yaml_file:
+            yaml_data = yaml.safe_load(hs_yaml_file)
+
+        default_formatid = yaml_data["store_metadata_namespace"]
+
     # Setup logging, create log file if it doesn't already exist
     hashstore_py_log = store_path + "/python_client.log"
     python_log_file_path = Path(hashstore_py_log)
@@ -768,6 +776,8 @@ def main():
     checksum_algorithm = getattr(args, "object_checksum_algorithm")
     size = getattr(args, "object_size")
     formatid = getattr(args, "object_formatid")
+    if formatid is None:
+        formatid = default_formatid
     knbvm_test = getattr(args, "knbvm_flag")
     # Instantiate HashStore Client
     props = parser.load_store_properties(store_path_config_yaml)

tests/test_filehashstore.py

@@ -1057,7 +1057,7 @@ def test_resolve_path_refs_pid(pids, store):
         resolved_pid_ref_abs_path = store._resolve_path("pid", pid)
         pid_refs_metadata_hashid = store._computehash(pid)
         calculated_pid_ref_path = (
-            store.refs + "/pid/" + "/".join(store._shard(pid_refs_metadata_hashid))
+            store.pids + "/" + "/".join(store._shard(pid_refs_metadata_hashid))
         )
 
         assert resolved_pid_ref_abs_path == calculated_pid_ref_path
@@ -1072,6 +1072,6 @@ def test_resolve_path_refs_cid(pids, store):
         cid = object_metadata.cid
 
         resolved_cid_ref_abs_path = store._resolve_path("cid", cid)
-        calculated_cid_ref_path = store.refs + "/cid/" + "/".join(store._shard(cid))
+        calculated_cid_ref_path = store.cids + "/" + "/".join(store._shard(cid))
 
         assert resolved_cid_ref_abs_path == calculated_cid_ref_path

tests/test_filehashstore_interface.py

@@ -1243,15 +1243,17 @@ def test_store_and_delete_objects_100_pids_1_cid(store):
     for i in range(1, upper_limit):
         pid_modified = f"dou.test.{str(i)}"
         store.store_object(pid_modified, path)
-    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/pid")]) == 100
-    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/cid")]) == 1
+    assert (
+        sum([len(files) for _, _, files in os.walk(store.root + "/refs/pids")]) == 100
+    )
+    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/cids")]) == 1
     assert store._count("objects") == 1
     # Delete
     for i in range(1, upper_limit):
         pid_modified = f"dou.test.{str(i)}"
         store.delete_object(pid_modified)
-    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/pid")]) == 0
-    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/cid")]) == 0
+    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/pids")]) == 0
+    assert sum([len(files) for _, _, files in os.walk(store.root + "/refs/cids")]) == 0
     assert store._count("objects") == 0