dcicutils 8.8.3.1b7__tar.gz → 8.8.3.1b8__tar.gz

{dcicutils-8.8.3.1b7 → dcicutils-8.8.3.1b8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.8.3.1b7
+Version: 8.8.3.1b8
 Summary: Utility package for interacting with the 4DN Data Portal and other 4DN resources
 Home-page: https://github.com/4dn-dcic/utils
 License: MIT

{dcicutils-8.8.3.1b7 → dcicutils-8.8.3.1b8}/dcicutils/structured_data.py

@@ -154,7 +154,32 @@ class StructuredDataSet:
 
     @property
     def resolved_refs_with_uuids(self) -> List[str]:
-        return list([{"path": resolved_ref[0], "uuid": resolved_ref[1]} for resolved_ref in self._resolved_refs])
+        return list([{"path": resolved_ref[0],
+                      "uuid": resolved_ref[1] if len(resolved_ref) >= 2 else None}
+                     for resolved_ref in self._resolved_refs])
+
+    @property
+    def unchecked_refs(self) -> List[str]:
+        """
+        Returns list of unchecked (for existence) references, grouped by reference path;
+        each object in the list has a path property and a srcs property which is a list of
+        src objects containing the type, column and row of the reference to the reference.
+        Note that this is only populated if the norefs option is specified.
+        """
+        def load_json(value: str) -> Optional[dict]:
+            try:
+                return json.loads(value)
+            except Exception:
+                return None
+        result = []
+        if self._norefs:
+            for ref in self._resolved_refs:
+                if len(ref) >= 3 and (ref_path := ref[0]) and (ref_src := load_json(ref[2])):
+                    if existing_ref := [item for item in result if item.get("path") == ref_path]:
+                        existing_ref[0]["srcs"].append(ref_src)
+                    else:
+                        result.append({"path": ref_path, "srcs": [ref_src]})
+        return result
 
     @property
     def upload_files(self) -> List[str]:
@@ -653,10 +678,14 @@ class Schema(SchemaBase):
         def map_ref(value: str, link_to: str, portal: Optional[Portal], src: Optional[str]) -> Any:
             nonlocal self, typeinfo
             if self._norefs:
+                # Here the caller has specified the (StructuredDataSet) norefs option
+                # which means we do not check for the existence of references at all.
                 if value:
-                    # Dump the src because this cannot add dictionary to a set; note that
-                    # this is ONLY used for smaht-submitr/submit-metadata-bundle --info --refs.
-                    self._resolved_refs.add((f"/{link_to}/{value}", json.dumps(src) if isinstance(src, dict) else None))
+                    # Dump the src as a JSON string because a dictionary cannot be added to a set; note
+                    # that this is ONLY used for smaht-submitr/submit-metadata-bundle --info --refs.
+                    # This info can be gotten at using StructureDataSet.unchecked_refs.
+                    self._resolved_refs.add((f"/{link_to}/{value}", None,
+                                             json.dumps(src) if isinstance(src, dict) else None))
                 return value
             if not value:
                 if (column := typeinfo.get("column")) and column in self.data.get("required", []):

{dcicutils-8.8.3.1b7 → dcicutils-8.8.3.1b8}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dcicutils"
-version = "8.8.3.1b7"  # TODO: To become 8.8.4
+version = "8.8.3.1b8"  # TODO: To become 8.8.4
 description = "Utility package for interacting with the 4DN Data Portal and other 4DN resources"
 authors = ["4DN-DCIC Team <support@4dnucleome.org>"]
 license = "MIT"