Release beta 15

security.txt

@@ -0,0 +1,2 @@
+If you'd like to report a security issue please contact us
+Contact: security-leads@data.humancellatlas.org

setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name="hca-metadata-api",
-    version="1.0b14",
+    version="1.0b15",
     license='MIT',
     install_requires=[
         'dataclasses >= 0.6'

src/humancellatlas/data/metadata/api.py

@@ -112,18 +112,31 @@ def __init__(self, entity: LinkedEntity, other_entity: Entity, forward: bool) ->
 
 @dataclass(frozen=True)
 class ProjectPublication:
-    publication_title: str
-    publication_url: Optional[str]
+    title: str
+    url: Optional[str]
 
     @classmethod
     def from_json(cls, json: JSON) -> 'ProjectPublication':
-        return cls(publication_title=json['publication_title'],
-                   publication_url=json.get('publication_url'))
+        title = lookup(json, 'title', 'publication_title')
+        url = lookup(json, 'url', 'publication_url', default=None)
+        return cls(title=title, url=url)
+
+    @property
+    def publication_title(self):
+        warnings.warn(f"ProjectPublication.publication_title is deprecated. "
+                      f"Use ProjectPublication.title instead.", DeprecationWarning)
+        return self.title
+
+    @property
+    def publication_url(self):
+        warnings.warn(f"ProjectPublication.publication_url is deprecated. "
+                      f"Use ProjectPublication.url instead.", DeprecationWarning)
+        return self.url
 
 
 @dataclass(frozen=True)
 class ProjectContact:
-    contact_name: str
+    name: str
     email: Optional[str]
     institution: Optional[str]  # optional up to project/5.3.0/contact
     laboratory: Optional[str]
@@ -132,12 +145,20 @@ class ProjectContact:
 
     @classmethod
     def from_json(cls, json: JSON) -> 'ProjectContact':
-        return cls(contact_name=json['contact_name'],
+        project_role = json.get('project_role')
+        project_role = ontology_label(project_role) if isinstance(project_role, dict) else project_role
+        return cls(name=lookup(json, 'name', 'contact_name'),
                    email=json.get('email'),
                    institution=json.get('institution'),
                    laboratory=json.get('laboratory'),
                    corresponding_contributor=json.get('corresponding_contributor'),
-                   project_role=json.get('project_role'))
+                   project_role=project_role)
+
+    @property
+    def contact_name(self) -> str:
+        warnings.warn(f"ProjectContact.contact_name is deprecated. "
+                      f"Use ProjectContact.name instead.", DeprecationWarning)
+        return self.name
 
 
 @dataclass(init=False)
@@ -290,24 +311,38 @@ def __init__(self, json: JSON) -> None:
 @dataclass(init=False)
 class CellSuspension(Biomaterial):
     estimated_cell_count: Optional[int]
-    selected_cell_type: Set[str]
+    selected_cell_types: Set[str]
 
     def __init__(self, json: JSON) -> None:
         super().__init__(json)
         content = json.get('content', json)
         self.estimated_cell_count = lookup(content, 'estimated_cell_count', 'total_estimated_cells', default=None)
-        self.selected_cell_type = {ontology_label(sct) for sct in content.get('selected_cell_type', [])}
+        self.selected_cell_types = {ontology_label(sct) for sct in
+                                    lookup(content, 'selected_cell_types', 'selected_cell_type', default=[])}
 
     @property
     def total_estimated_cells(self) -> int:
         warnings.warn(f"CellSuspension.total_estimated_cells is deprecated. "
                       f"Use CellSuspension.estimated_cell_count instead.", DeprecationWarning)
         return self.estimated_cell_count
 
+    @property
+    def selected_cell_type(self) -> Set[str]:
+        warnings.warn(f"CellSuspension.selected_cell_type is deprecated. "
+                      f"Use CellSuspension.selected_cell_types instead.", DeprecationWarning)
+        return self.selected_cell_types
+
 
 @dataclass(init=False)
 class CellLine(Biomaterial):
-    pass
+    cell_line_type: str
+    model_organ: Optional[str]
+
+    def __init__(self, json: JSON) -> None:
+        super().__init__(json)
+        content = json.get('content', json)
+        self.cell_line_type = content['cell_line_type']
+        self.model_organ = ontology_label(content.get('model_organ'), default=None)
 
 
 @dataclass(init=False)
@@ -527,7 +562,7 @@ def from_json(cls, json: JSON):
 
 @dataclass(init=False)
 class File(LinkedEntity):
-    file_format: str
+    format: str
     from_processes: MutableMapping[UUID4, Process] = field(repr=False)
     to_processes: MutableMapping[UUID4, Process]
     manifest_entry: ManifestEntry
@@ -536,7 +571,7 @@ def __init__(self, json: JSON, manifest: Mapping[str, ManifestEntry]):
         super().__init__(json)
         content = json.get('content', json)
         core = content['file_core']
-        self.file_format = core['file_format']
+        self.format = lookup(core, 'format', 'file_format')
         self.manifest_entry = manifest[core['file_name']]
         self.from_processes = {}
         self.to_processes = {}
@@ -550,6 +585,12 @@ def _connect_to(self, other: Entity, forward: bool) -> None:
         else:
             raise LinkError(self, other, forward)
 
+    @property
+    def file_format(self) -> str:
+        warnings.warn(f"File.file_format is deprecated. "
+                      f"Use File.format instead.", DeprecationWarning)
+        return self.format
+
 
 @dataclass(init=False)
 class SequenceFile(File):
@@ -602,14 +643,20 @@ def from_json(cls, json: JSON) -> Iterable['Link']:
             # vx
             process_id = UUID4(json['process'])
             for source_id in json['inputs']:
-                yield cls(source_id=UUID4(source_id), source_type=json['input_type'],
-                          destination_id=process_id, destination_type='process')
+                yield cls(source_id=UUID4(source_id),
+                          source_type=json['input_type'],
+                          destination_id=process_id,
+                          destination_type='process')
             for destination_id in json['outputs']:
-                yield cls(source_id=process_id, source_type='process',
-                          destination_id=UUID4(destination_id), destination_type=json['output_type'])
+                yield cls(source_id=process_id,
+                          source_type='process',
+                          destination_id=UUID4(destination_id),
+                          destination_type=json['output_type'])
             for protocol in json['protocols']:
-                yield cls(source_id=process_id, source_type='process',
-                          destination_id=UUID4(protocol['protocol_id']), destination_type=protocol['protocol_type'])
+                yield cls(source_id=process_id,
+                          source_type='process',
+                          destination_id=UUID4(protocol['protocol_id']),
+                          destination_type=lookup(protocol, 'type', 'protocol_type'))
 
 
 @dataclass(init=False)

src/humancellatlas/data/metadata/helpers/dss.py

@@ -76,7 +76,13 @@ def download_file(item):
         file_version = manifest_entry['version']
         logger.debug("Getting file '%s' (%s.%s) from DSS.", file_name, file_uuid, file_version)
         # noinspection PyUnresolvedReferences
-        file_contents = client.get_file(uuid=file_uuid, version=file_version, replica='aws')
+        file_contents = client.get_file(uuid=file_uuid, version=file_version, replica=replica)
+
+        # Work around https://github.com/HumanCellAtlas/data-store/issues/2073
+        if replica == 'gcp' and isinstance(file_contents, bytes):  # pragma: no cover
+            import json
+            file_contents = json.loads(file_contents)
+
         if not isinstance(file_contents, dict):
             raise TypeError(f'Expecting file {file_uuid}.{file_version} '
                             f'to contain a JSON object ({dict}), '
@@ -92,14 +98,13 @@ def download_file(item):
     return bundle['version'], manifest, dict(metadata_files)
 
 
-def dss_client(deployment: Optional[str] = None) -> DSSClient:
+def dss_client(deployment: str = 'prod') -> DSSClient:
     """
     Return a DSS client to DSS production or the specified DSS deployment.
 
-    :param deployment: The name of a DSS deployment like `dev`, `integration` or `staging`. If None, the production
-                       deployment (`prod`) will be used.
+    :param deployment: The name of a DSS deployment like `dev`, `integration`, `staging` or `prod`.
     """
-    deployment = deployment + "." if deployment else ""
+    deployment = "" if deployment == "prod" else deployment + "."
     swagger_url = f'https://dss.{deployment}data.humancellatlas.org/v1/swagger.json'
     client = DSSClient(swagger_url=swagger_url)
     client.timeout_policy = Timeout(connect=10, read=40)