Merge pull request #243 from ihmeuw/collijk/bugfix/clean-up-artifact-…

…test-warnings-clean

Squash warnings in artifact generation code

src/vivarium/framework/artifact/artifact.py

@@ -44,7 +44,7 @@ def __init__(self, path: Union[str, Path], filter_terms: List[str] = None):
         self._draw_column_filter = _parse_draw_filters(filter_terms)
         self._cache = {}
 
-        self.create_hdf_with_keyspace()
+        self.create_hdf_with_keyspace(self._path)
         self._keys = Keys(self._path)
 
     @property
@@ -62,12 +62,24 @@ def filter_terms(self) -> List[str]:
         """Filters that will be applied to the requested data on loads."""
         return self._filter_terms
 
-    def create_hdf_with_keyspace(self):
+    @staticmethod
+    def create_hdf_with_keyspace(path: Path):
         """Creates the artifact HDF file and adds a node to track keys."""
-        if not self._path.is_file():
-            warnings.warn(f"No artifact found at {self._path}. Building new artifact.")
-            hdf.touch(self._path)
-            hdf.write(self._path, "metadata.keyspace", ["metadata.keyspace"])
+        if not path.is_file():
+            warnings.warn(f"No artifact found at {path}. Building new artifact.")
+            hdf.touch(path)
+
+        keys = hdf.get_keys(path)
+        if keys and "metadata.keyspace" not in keys:
+            raise ArtifactException(
+                "Attempting to construct an Artifact from a malformed existing file. "
+                "This can occur when constructing an Artifact from an existing file when "
+                "the existing file was generated by some other hdf writing mechanism "
+                "(e.g. pd.to_hdf) rather than generating the the file using this class "
+                "and a non-existent or empty hdf file."
+            )
+        if not keys:
+            hdf.write(path, "metadata.keyspace", ["metadata.keyspace"])
 
     def load(self, entity_key: str) -> Any:
         """Loads the data associated with provided entity_key.

src/vivarium/framework/artifact/hdf.py

@@ -196,7 +196,7 @@ def remove(path: Union[str, Path], entity_key: str):
         file.remove_node(entity_key.path, recursive=True)
 
 
-def get_keys(path: str) -> List[str]:
+def get_keys(path: Union[str, Path]) -> List[str]:
     """Gets key representation of all paths in an HDF file.
 
     Parameters
@@ -426,7 +426,7 @@ def _get_valid_filter_terms(filter_terms, colnames):
         # then split each condition out
         t = re.split("[&|]", t)
         # get the unique columns referenced by this term
-        term_columns = set([re.split("[<=>\s]", i.strip())[0] for i in t])
+        term_columns = set([re.split(r"[<=>\s]", i.strip())[0] for i in t])
         if not term_columns.issubset(colnames):
             valid_terms.remove(term)
     return valid_terms if valid_terms else None