[p] Extract is_stitched from manifest entries (partial #6299)

attic/scripts/recan_bundle_tdr.py

@@ -177,7 +177,6 @@ def dss_bundle_to_tdr(bundle: Bundle, source: TDRSourceRef) -> TDRHCABundle:
     links_entry = None
     for entry in manifest:
         entry['version'] = convert_version(entry['version'])
-        entry['is_stitched'] = False
         if entry['name'] == 'links.json':
             links_entry = entry
         if entry['indexed']:

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

@@ -81,7 +81,8 @@ def transformers(self,
                                 version=bundle.version,
                                 manifest=bundle.manifest,
                                 metadata_files=bundle.metadata_files,
-                                links_json=bundle.links)
+                                links_json=bundle.links,
+                                stitched_entity_ids=bundle.stitched)
 
         def transformers():
             for transformer_cls in self.transformer_types():

src/azul/plugins/metadata/hca/bundle.py

@@ -41,6 +41,7 @@ class HCABundle(Bundle[BUNDLE_FQID], ABC):
     """
     metadata_files: MutableJSON
     links: MutableJSON
+    stitched: set[str] = attrs.field(factory=set)
 
     def reject_joiner(self, catalog: CatalogName):
         self._reject_joiner(self.manifest)
@@ -52,14 +53,21 @@ def to_json(self) -> MutableJSON:
             'manifest': self.manifest,
             'metadata': self.metadata_files,
             'links': self.links,
+            'stitched': sorted(self.stitched)
         }
 
     @classmethod
     def from_json(cls, fqid: BUNDLE_FQID, json_: JSON) -> 'Bundle':
         manifest = json_['manifest']
         metadata = json_['metadata']
         links = json_['links']
+        stitched = json_['stitched']
         assert isinstance(manifest, list), manifest
         assert isinstance(metadata, dict), metadata
         assert isinstance(links, dict), links
-        return cls(fqid=fqid, manifest=manifest, metadata_files=metadata, links=links)
+        assert isinstance(stitched, list), stitched
+        return cls(fqid=fqid,
+                   manifest=manifest,
+                   metadata_files=metadata,
+                   links=links,
+                   stitched=set(stitched))

src/azul/plugins/metadata/hca/indexer/transform.py

@@ -1428,7 +1428,7 @@ def entity_type(cls) -> str:
         return 'files'
 
     def _entities(self) -> Iterable[api.File]:
-        return api.not_stitched(self.api_bundle.files.values())
+        return self.api_bundle.not_stitched(self.api_bundle.files)
 
     def _transform(self, files: Iterable[api.File]) -> Iterable[Contribution]:
         zarr_stores: Mapping[str, list[api.File]] = self.group_zarrs(files)
@@ -1586,7 +1586,7 @@ def inner_entity_types(cls) -> frozenset[str]:
 
     def _entities(self) -> Iterable[Sample]:
         samples: dict[str, Sample] = dict()
-        for file in api.not_stitched(self.api_bundle.files.values()):
+        for file in self.api_bundle.not_stitched(self.api_bundle.files):
             self._find_ancestor_samples(file, samples)
         return samples.values()
 
@@ -1643,7 +1643,7 @@ def _singleton_entity(self) -> DatedEntity:
         raise NotImplementedError
 
     def _dated_entities(self) -> Iterable[DatedEntity]:
-        return api.not_stitched(self.api_bundle.entities.values())
+        return self.api_bundle.not_stitched(self.api_bundle.entities)
 
     def estimate(self, partition: BundlePartition) -> int:
         return int(partition.contains(self._singleton_id))

src/azul/plugins/repository/tdr_hca/__init__.py

@@ -189,13 +189,14 @@ def add_entity(self,
                    row: BigQueryRow,
                    is_stitched: bool
                    ) -> None:
+        if is_stitched:
+            self.stitched.add(entity.entity_id)
         self._add_manifest_entry(name=entity_key,
                                  uuid=entity.entity_id,
                                  version=TDRPlugin.format_version(row['version']),
                                  size=row['content_size'],
                                  content_type='application/json',
-                                 dcp_type=f'"metadata/{row["schema_type"]}"',
-                                 is_stitched=is_stitched)
+                                 dcp_type=f'"metadata/{row["schema_type"]}"')
         if entity.entity_type.endswith('_file'):
             descriptor = json.loads(row['descriptor'])
             self._add_manifest_entry(name=row['file_name'],
@@ -204,7 +205,6 @@ def add_entity(self,
                                      size=descriptor['size'],
                                      content_type=descriptor['content_type'],
                                      dcp_type='data',
-                                     is_stitched=is_stitched,
                                      checksums=Checksums.from_json(descriptor),
                                      drs_uri=self._parse_drs_uri(row['file_id'], descriptor))
         content = row['content']
@@ -241,7 +241,6 @@ def _add_manifest_entry(self,
                             size: int,
                             content_type: str,
                             dcp_type: str,
-                            is_stitched: bool,
                             checksums: Optional[Checksums] = None,
                             drs_uri: Optional[str] = None) -> None:
 
@@ -251,7 +250,6 @@ def _add_manifest_entry(self,
             'version': version,
             'content-type': f'{content_type}; dcp-type={dcp_type}',
             'size': size,
-            'is_stitched': is_stitched,
             **(
                 {
                     'indexed': True,

src/humancellatlas/data/metadata/api.py

@@ -18,8 +18,8 @@
     chain,
 )
 from typing import (
+    AbstractSet,
     Iterable,
-    Iterator,
     Mapping,
     MutableMapping,
     Optional,
@@ -77,7 +77,6 @@ class ManifestEntry:
     # FIXME: Change Bundle.version and ManifestEntry.version from string to datetime
     #        https://github.com/DataBiosphere/hca-metadata-api/issues/48
     version: str
-    is_stitched: bool = field(init=False)
 
     def __init__(self, json: MutableJSON):
         # '/' was once forbidden in file paths and was encoded with '!'. Now
@@ -87,7 +86,6 @@ def __init__(self, json: MutableJSON):
         self.json = json
         self.content_type = json['content-type']
         self.uuid = UUID4(json['uuid'])
-        self.is_stitched = json.get('is_stitched', False)
         for f in fields(self):
             if f.init:
                 value = json.get(f.name)
@@ -106,13 +104,6 @@ class Entity:
     submission_date: datetime
     update_date: Optional[datetime]
 
-    @property
-    def is_stitched(self):
-        if self.metadata_manifest_entry is None:
-            return False
-        else:
-            return self.metadata_manifest_entry.is_stitched
-
     @classmethod
     def from_json(cls,
                   json: JSON,
@@ -170,20 +161,6 @@ def accept(self, visitor: 'EntityVisitor') -> None:
 E = TypeVar('E', bound=Entity)
 
 
-# noinspection PyPep8Naming
-@dataclass(frozen=True)
-class not_stitched(Iterable[E]):
-    """
-    An iterable of the entities in the argument iterable that are not stitched.
-    This is an iterable, so it can be consumed repeatedly.
-    """
-
-    entities: Iterable[E]
-
-    def __iter__(self) -> Iterator[E]:
-        return (e for e in self.entities if not e.is_stitched)
-
-
 class TypeLookupError(Exception):
 
     def __init__(self, described_by: str) -> None:
@@ -1001,10 +978,12 @@ def __init__(self,
                  version: str,
                  manifest: MutableJSONs,
                  metadata_files: Mapping[str, JSON],
-                 links_json: JSON):
+                 links_json: JSON,
+                 stitched_entity_ids: AbstractSet[str] = frozenset()):
         self.uuid = UUID4(uuid)
         self.version = version
         self.manifest = {m.name: m for m in map(ManifestEntry, manifest)}
+        self.stitched = stitched_entity_ids
 
         json_by_core_cls: MutableMapping[type[E], list[tuple[JSON, ManifestEntry]]] = defaultdict(list)
         for file_name, json in metadata_files.items():
@@ -1098,6 +1077,13 @@ def visit(entity: Entity) -> Mapping[UUID4, L]:
 
         return recurse(self.root_entities().values())
 
+    def not_stitched(self, entities: Mapping[UUID, E]) -> list[E]:
+        return [
+            entity
+            for uuid, entity in entities.items()
+            if str(uuid) not in self.stitched
+        ]
+
     @cached_property
     def leaf_cell_suspensions(self) -> Mapping[UUID4, CellSuspension]:
         return self.leaf_entities(CellSuspension)