Push down HCA specifics to a Bundle subclass (#4940)

src/azul/indexer/__init__.py

@@ -30,7 +30,6 @@
 from azul.types import (
     JSON,
     MutableJSON,
-    MutableJSONs,
     SupportsLessThan,
     get_generic_type_params,
 )
@@ -396,23 +395,6 @@ def to_json(self) -> SourcedBundleFQIDJSON:
 @attr.s(auto_attribs=True, kw_only=True)
 class Bundle(ABC, Generic[BUNDLE_FQID]):
     fqid: BUNDLE_FQID
-    manifest: MutableJSONs
-    """
-    Each item of the `manifest` attribute's value has this shape:
-    {
-        'content-type': 'application/json; dcp-type="metadata/biomaterial"',
-        'crc32c': 'fd239631',
-        'indexed': True,
-        'name': 'cell_suspension_0.json',
-        's3_etag': 'aa31c093cc816edb1f3a42e577872ec6',
-        'sha1': 'f413a9a7923dee616309e4f40752859195798a5d',
-        'sha256': 'ea4c9ed9e53a3aa2ca4b7dffcacb6bbe9108a460e8e15d2b3d5e8e5261fb043e',
-        'size': 1366,
-        'uuid': '0136ebb4-1317-42a0-8826-502fae25c29f',
-        'version': '2019-05-16T162155.020000Z'
-    }
-    """
-    metadata_files: MutableJSON
 
     @property
     def uuid(self) -> BundleUUID:
@@ -432,6 +414,10 @@ def drs_path(self, manifest_entry: JSON) -> Optional[str]:
         """
         raise NotImplementedError
 
+    @abstractmethod
+    def to_json(self) -> MutableJSON:
+        raise NotImplementedError
+
 
 class BundlePartition(UUIDPartition['BundlePartition']):
     """

src/azul/plugins/anvil.py

@@ -0,0 +1,89 @@
+from abc import (
+    ABC,
+)
+from typing import (
+    AbstractSet,
+    Generic,
+    Iterable,
+    Optional,
+    TypeVar,
+    Union,
+)
+
+import attr
+from more_itertools import (
+    one,
+)
+
+from azul.indexer import (
+    BUNDLE_FQID,
+    Bundle,
+)
+from azul.indexer.document import (
+    EntityReference,
+    EntityType,
+)
+from azul.types import (
+    MutableJSON,
+)
+
+# AnVIL snapshots do not use UUIDs for primary/foreign keys.
+# This type alias helps us distinguish these keys from the document UUIDs,
+# which are drawn from the `datarepo_row_id` column.
+# Note that entities from different tables may have the same key, so
+# `KeyReference` should be used when mixing keys from different entity types.
+Key = str
+
+
+@attr.s(frozen=True, auto_attribs=True, kw_only=True, slots=True)
+class KeyReference:
+    key: Key
+    entity_type: EntityType
+
+
+ENTITY_REF = TypeVar(name='ENTITY_REF', bound=Union[EntityReference, KeyReference])
+
+
+@attr.s(auto_attribs=True, frozen=True, kw_only=True, order=False)
+class Link(Generic[ENTITY_REF]):
+    inputs: AbstractSet[ENTITY_REF] = attr.ib(factory=frozenset)
+    activity: Optional[ENTITY_REF] = attr.ib(default=None)
+    outputs: AbstractSet[ENTITY_REF] = attr.ib(factory=frozenset)
+
+    @property
+    def all_entities(self) -> AbstractSet[ENTITY_REF]:
+        return self.inputs | self.outputs | (set() if self.activity is None else {self.activity})
+
+    def to_json(self) -> MutableJSON:
+        return {
+            'inputs': sorted(map(str, self.inputs)),
+            'activity': str(self.activity),
+            'outputs': sorted(map(str, self.outputs))
+        }
+
+    @classmethod
+    def merge(cls, links: Iterable['Link']) -> 'Link':
+        return cls(inputs=frozenset.union(*[link.inputs for link in links]),
+                   activity=one({link.activity for link in links}),
+                   outputs=frozenset.union(*[link.outputs for link in links]))
+
+    def __lt__(self, other: 'Link') -> bool:
+        if self.activity is None or other.activity is None:
+            return False
+        else:
+            return self.activity < other.activity
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class AnvilBundle(Bundle[BUNDLE_FQID], ABC):
+    entities: dict[EntityReference, MutableJSON] = attr.ib(factory=dict)
+    links: set[Link[EntityReference]] = attr.ib(factory=set)
+
+    def to_json(self) -> MutableJSON:
+        return {
+            'entities': {
+                str(entity_ref): entity
+                for entity_ref, entity in sorted(self.entities.items())
+            },
+            'links': [link.to_json() for link in sorted(self.links)]
+        }

src/azul/plugins/hca.py

@@ -0,0 +1,41 @@
+from abc import (
+    ABC,
+)
+
+import attr
+
+from azul.indexer import (
+    BUNDLE_FQID,
+    Bundle,
+)
+from azul.types import (
+    MutableJSON,
+    MutableJSONs,
+)
+
+
+@attr.s(auto_attribs=True, kw_only=True)
+class HCABundle(Bundle[BUNDLE_FQID], ABC):
+    manifest: MutableJSONs
+    """
+    Each item of the `manifest` attribute's value has this shape:
+    {
+        'content-type': 'application/json; dcp-type="metadata/biomaterial"',
+        'crc32c': 'fd239631',
+        'indexed': True,
+        'name': 'cell_suspension_0.json',
+        's3_etag': 'aa31c093cc816edb1f3a42e577872ec6',
+        'sha1': 'f413a9a7923dee616309e4f40752859195798a5d',
+        'sha256': 'ea4c9ed9e53a3aa2ca4b7dffcacb6bbe9108a460e8e15d2b3d5e8e5261fb043e',
+        'size': 1366,
+        'uuid': '0136ebb4-1317-42a0-8826-502fae25c29f',
+        'version': '2019-05-16T162155.020000Z'
+    }
+    """
+    metadata_files: MutableJSON
+
+    def to_json(self) -> MutableJSON:
+        return {
+            'manifest': self.manifest,
+            'metadata': self.metadata_files
+        }

src/azul/plugins/metadata/anvil/__init__.py

@@ -6,15 +6,15 @@
     Type,
 )
 
-from azul.indexer import (
-    Bundle,
-)
 from azul.plugins import (
     DocumentSlice,
     ManifestConfig,
     MetadataPlugin,
     Sorting,
 )
+from azul.plugins.anvil import (
+    AnvilBundle,
+)
 from azul.plugins.metadata.anvil.indexer.transform import (
     ActivityTransformer,
     BaseTransformer,
@@ -42,7 +42,7 @@
 )
 
 
-class Plugin(MetadataPlugin[Bundle]):
+class Plugin(MetadataPlugin[AnvilBundle]):
 
     @classmethod
     def atlas(cls) -> str:
@@ -71,7 +71,7 @@ def transformer_types(self) -> Iterable[Type[BaseTransformer]]:
             FileTransformer,
         )
 
-    def transformers(self, bundle: Bundle, *, delete: bool) -> Iterable[BaseTransformer]:
+    def transformers(self, bundle: AnvilBundle, *, delete: bool) -> Iterable[BaseTransformer]:
         return [
             transformer_cls(bundle=bundle, deleted=delete)
             for transformer_cls in self.transformer_types()