dcicutils 8.8.6.1b1__tar.gz → 8.8.6.1b3__tar.gz

{dcicutils-8.8.6.1b1 → dcicutils-8.8.6.1b3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.8.6.1b1
+Version: 8.8.6.1b3
 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-8.8.6.1b1 → dcicutils-8.8.6.1b3}/dcicutils/structured_data.py

@@ -351,18 +351,23 @@ class StructuredDataSet:
 
     def _load_json_file(self, file: str) -> None:
         with open(file) as f:
-            file_json = json.load(f)
-            schema_inferred_from_file_name = Schema.type_name(file)
-            if self._portal.get_schema(schema_inferred_from_file_name) is not None:
+            item = json.load(f)
+            if ((schema_name_inferred_from_file_name := Schema.type_name(file)) and
+                (self._portal.get_schema(schema_name_inferred_from_file_name) is not None)):  # noqa
                 # If the JSON file name looks like a schema name then assume it
                 # contains an object or an array of object of that schema type.
-                self._add(Schema.type_name(file), file_json)
-            elif isinstance(file_json, dict):
+                if self._merge:
+                    item = self._merge_with_existing_portal_object(item, schema_name_inferred_from_file_name)
+                self._add(Schema.type_name(file), item)
+            elif isinstance(item, dict):
                 # Otherwise if the JSON file name does not look like a schema name then
                 # assume it a dictionary where each property is the name of a schema, and
                 # which (each property) contains a list of object of that schema type.
-                for schema_name in file_json:
-                    self._add(schema_name, file_json[schema_name])
+                for schema_name in item:
+                    item = item[schema_name]
+                    if self._merge:
+                        item = self._merge_with_existing_portal_object(item, schema_name)
+                    self._add(schema_name, item)
 
     def _load_reader(self, reader: RowReader, type_name: str) -> None:
         schema = None
@@ -386,14 +391,12 @@ class StructuredDataSet:
                     self._add_properties(structured_row, self._autoadd_properties, schema)
             # New merge functionality (2024-05-25).
             if self._merge:
-                for identifying_path in self.get_identifying_paths(self._portal, structured_row, type_name):
-                    if existing_portal_object := self._portal.get_metadata(identifying_path):
-                        structured_row = merge_objects(existing_portal_object, structured_row)
+                structured_row = self._merge_with_existing_portal_object(structured_row, schema_name)
             if (prune_error := self._prune_structured_row(structured_row)) is not None:
                 self._note_error({"src": create_dict(type=schema_name, row=reader.row_number),
                                   "error": prune_error}, "validation")
             else:
-                self._add(type_name, structured_row)
+                self._add(type_name, structured_row)  # TODO: why type_name and not schema_name?
             if self._progress:
                 self._progress({
                     PROGRESS.LOAD_ITEM: self._nrows,
@@ -434,6 +437,18 @@ class StructuredDataSet:
             if name not in structured_row and (not schema or schema.data.get("properties", {}).get(name)):
                 structured_row[name] = properties[name]
 
+    def _merge_with_existing_portal_object(self, portal_object: dict, portal_type: str) -> dict:
+        """
+        Given a Portal object (presumably/in-practice from the given metadata), if there is
+        an existing Portal item, identified by the identifying properties for the given object,
+        then merges the given object into the existing one and returns the result; otherwise
+        just returns the given object. Note that the given object may be CHANGED in place.
+        """
+        for identifying_path in self._portal.get_identifying_paths(portal_object, portal_type):
+            if existing_portal_object := self._portal.get_metadata(identifying_path, raw=True, raise_exception=False):
+                return merge_objects(existing_portal_object, portal_object)
+        return portal_object
+
     def _is_ref_lookup_specified_type(ref_lookup_flags: int) -> bool:
         return (ref_lookup_flags &
                 Portal.LOOKUP_SPECIFIED_TYPE) == Portal.LOOKUP_SPECIFIED_TYPE

{dcicutils-8.8.6.1b1 → dcicutils-8.8.6.1b3}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "dcicutils"
-version = "8.8.6.1b1"
+version = "8.8.6.1b3"
 description = "Utility package for interacting with the 4DN Data Portal and other 4DN resources"
 authors = ["4DN-DCIC Team <support@4dnucleome.org>"]
 license = "MIT"

{dcicutils-8.8.6.1b1 → dcicutils-8.8.6.1b3}/dcicutils/portal_utils.py

@@ -418,31 +418,6 @@ class Portal:
             return []
         return schemas_super_type_map.get(type_name, [])
 
-    def url(self, url: str, raw: bool = False, database: bool = False) -> str:
-        if not isinstance(url, str) or not url:
-            return "/"
-        elif (lowercase_url := url.lower()).startswith("http://") or lowercase_url.startswith("https://"):
-            return url
-        elif not (url := re.sub(r"/+", "/", url)).startswith("/"):
-            url = "/"
-        url = self.server + url if self.server else url
-        if isinstance(raw, bool) and raw:
-            url += ("&" if "?" in url else "?") + "frame=raw"
-        if isinstance(database, bool) and database:
-            url += ("&" if "?" in url else "?") + "datastore=database"
-        return url
-
-    def _kwargs(self, **kwargs) -> dict:
-        if "headers" in kwargs:
-            result_kwargs = {"headers": kwargs["headers"]}
-        else:
-            result_kwargs = {"headers": {"Content-type": Portal.MIME_TYPE_JSON, "Accept": Portal.MIME_TYPE_JSON}}
-        if self.key_pair:
-            result_kwargs["auth"] = self.key_pair
-        if isinstance(timeout := kwargs.get("timeout"), int):
-            result_kwargs["timeout"] = timeout
-        return result_kwargs
-
     @function_cache(maxsize=100, serialize_key=True)
     def get_identifying_paths(self, portal_object: dict, portal_type: Optional[str] = None) -> List[str]:
         """
@@ -491,6 +466,31 @@ class Portal:
                 identifying_properties.insert(0, favored_identifying_property)
         return identifying_properties
 
+    def url(self, url: str, raw: bool = False, database: bool = False) -> str:
+        if not isinstance(url, str) or not url:
+            return "/"
+        elif (lowercase_url := url.lower()).startswith("http://") or lowercase_url.startswith("https://"):
+            return url
+        elif not (url := re.sub(r"/+", "/", url)).startswith("/"):
+            url = "/"
+        url = self.server + url if self.server else url
+        if isinstance(raw, bool) and raw:
+            url += ("&" if "?" in url else "?") + "frame=raw"
+        if isinstance(database, bool) and database:
+            url += ("&" if "?" in url else "?") + "datastore=database"
+        return url
+
+    def _kwargs(self, **kwargs) -> dict:
+        if "headers" in kwargs:
+            result_kwargs = {"headers": kwargs["headers"]}
+        else:
+            result_kwargs = {"headers": {"Content-type": Portal.MIME_TYPE_JSON, "Accept": Portal.MIME_TYPE_JSON}}
+        if self.key_pair:
+            result_kwargs["auth"] = self.key_pair
+        if isinstance(timeout := kwargs.get("timeout"), int):
+            result_kwargs["timeout"] = timeout
+        return result_kwargs
+
     @staticmethod
     def _default_keys_file(app: Optional[str], env: Optional[str], server: Optional[str]) -> Optional[str]:
         def infer_app_from_env(env: str) -> Optional[str]:  # noqa
@@ -566,6 +566,22 @@ class Portal:
             response = TestResponseWrapper(response)
         return response
 
+    @staticmethod
+    def _create_vapp(arg: Union[TestApp, VirtualApp, PyramidRouter, str] = None) -> TestApp:
+        if isinstance(arg, TestApp):
+            return arg
+        elif isinstance(arg, VirtualApp):
+            if not isinstance(arg.wrapped_app, TestApp):
+                raise Exception("Portal._create_vapp VirtualApp argument error.")
+            return arg.wrapped_app
+        if isinstance(arg, PyramidRouter):
+            router = arg
+        elif isinstance(arg, str) or not arg:
+            router = pyramid_get_app(arg or "development.ini", "app")
+        else:
+            raise Exception("Portal._create_vapp argument error.")
+        return TestApp(router, {"HTTP_ACCEPT": Portal.MIME_TYPE_JSON, "REMOTE_USER": "TEST"})
+
     @staticmethod
     def create_for_testing(arg: Optional[Union[str, bool, List[dict], dict, Callable]] = None) -> Portal:
         if isinstance(arg, list) or isinstance(arg, dict) or isinstance(arg, Callable):
@@ -597,22 +613,6 @@ class Portal:
         with temporary_file(content=minimal_ini_for_testing, suffix=".ini") as ini_file:
             return Portal(ini_file)
 
-    @staticmethod
-    def _create_vapp(arg: Union[TestApp, VirtualApp, PyramidRouter, str] = None) -> TestApp:
-        if isinstance(arg, TestApp):
-            return arg
-        elif isinstance(arg, VirtualApp):
-            if not isinstance(arg.wrapped_app, TestApp):
-                raise Exception("Portal._create_vapp VirtualApp argument error.")
-            return arg.wrapped_app
-        if isinstance(arg, PyramidRouter):
-            router = arg
-        elif isinstance(arg, str) or not arg:
-            router = pyramid_get_app(arg or "development.ini", "app")
-        else:
-            raise Exception("Portal._create_vapp argument error.")
-        return TestApp(router, {"HTTP_ACCEPT": Portal.MIME_TYPE_JSON, "REMOTE_USER": "TEST"})
-
     @staticmethod
     def _create_router_for_testing(endpoints: Optional[List[Dict[str, Union[str, Callable]]]] = None) -> PyramidRouter:
         if isinstance(endpoints, dict):