Blob metadata (#1154)

* Add blob metadata

* fix

* change base config class

* make backward-compatible

src/evidently/ui/app.py

@@ -4,15 +4,15 @@
 
 from evidently._pydantic_compat import SecretStr
 from evidently.ui.components.base import AppBuilder
-from evidently.ui.config import Config
+from evidently.ui.config import AppConfig
 from evidently.ui.config import load_config
 from evidently.ui.config import settings
 from evidently.ui.local_service import LocalConfig
 from evidently.ui.security.token import TokenSecurityComponent
 from evidently.ui.storage.common import EVIDENTLY_SECRET_ENV
 
 
-def create_app(config: Config):
+def create_app(config: AppConfig):
     with config.context() as ctx:
         builder = AppBuilder(ctx)
         ctx.apply(builder)
@@ -21,7 +21,7 @@ def create_app(config: Config):
         return app
 
 
-def run(config: Config):
+def run(config: AppConfig):
     app = create_app(config)
     uvicorn.run(app, host=config.service.host, port=config.service.port)
 

src/evidently/ui/base.py

@@ -52,14 +52,19 @@
 from evidently.utils.dashboard import TemplateParams
 
 
+class BlobMetadata(BaseModel):
+    id: BlobID
+    size: Optional[int]
+
+
 class SnapshotMetadata(BaseModel):
     id: UUID4
     name: Optional[str] = None
     timestamp: datetime.datetime
     metadata: Dict[str, MetadataValueType]
     tags: List[str]
     is_report: bool
-    blob_id: BlobID
+    blob: "BlobMetadata"
 
     _project: "Project" = PrivateAttr(None)
     _dashboard_info: "DashboardInfo" = PrivateAttr(None)
@@ -81,15 +86,15 @@ def bind(self, project: "Project"):
         return self
 
     @classmethod
-    def from_snapshot(cls, snapshot: Snapshot, blob_id: BlobID) -> "SnapshotMetadata":
+    def from_snapshot(cls, snapshot: Snapshot, blob: "BlobMetadata") -> "SnapshotMetadata":
         return SnapshotMetadata(
             id=snapshot.id,
             name=snapshot.name,
             timestamp=snapshot.timestamp,
             metadata=snapshot.metadata,
             tags=snapshot.tags,
             is_report=snapshot.is_report,
-            blob_id=blob_id,
+            blob=blob,
         )
 
     @property
@@ -250,7 +255,7 @@ def list_projects(self, project_ids: Optional[Set[ProjectID]]) -> List[Project]:
         raise NotImplementedError
 
     @abstractmethod
-    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob_id: str):
+    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob: "BlobMetadata"):
         raise NotImplementedError
 
     @abstractmethod
@@ -286,16 +291,19 @@ class BlobStorage(ABC):
     def open_blob(self, id: BlobID):
         raise NotImplementedError
 
-    def put_blob(self, path: str, obj):
+    def put_blob(self, blob_id: str, obj):
         raise NotImplementedError
 
     def get_snapshot_blob_id(self, project_id: ProjectID, snapshot: Snapshot) -> BlobID:
         raise NotImplementedError
 
-    def put_snapshot(self, project_id: ProjectID, snapshot: Snapshot) -> BlobID:
+    def put_snapshot(self, project_id: ProjectID, snapshot: Snapshot) -> BlobMetadata:
         id = self.get_snapshot_blob_id(project_id, snapshot)
         self.put_blob(id, json.dumps(snapshot.dict(), cls=NumpyEncoder))
-        return id
+        return self.get_blob_metadata(id)
+
+    def get_blob_metadata(self, blob_id: BlobID) -> BlobMetadata:
+        raise NotImplementedError
 
 
 class DataStorage(ABC):
@@ -696,8 +704,8 @@ def add_snapshot(self, user_id: UserID, project_id: ProjectID, snapshot: Snapsho
             user.id, EntityType.Project, project_id, Permission.PROJECT_SNAPSHOT_ADD
         ):
             raise ProjectNotFound()  # todo: better exception
-        blob_id = self.blob.put_snapshot(project_id, snapshot)
-        self.metadata.add_snapshot(project_id, snapshot, blob_id)
+        blob = self.blob.put_snapshot(project_id, snapshot)
+        self.metadata.add_snapshot(project_id, snapshot, blob)
         self.data.extract_points(project_id, snapshot)
 
     def delete_snapshot(self, user_id: UserID, project_id: ProjectID, snapshot_id: SnapshotID):
@@ -733,7 +741,7 @@ def load_snapshot(
     ) -> Snapshot:
         if isinstance(snapshot, SnapshotID):
             snapshot = self.get_snapshot_metadata(user_id, project_id, snapshot)
-        with self.blob.open_blob(snapshot.blob_id) as f:
+        with self.blob.open_blob(snapshot.blob.id) as f:
             return parse_obj_as(Snapshot, json.load(f))
 
     def get_snapshot_metadata(

src/evidently/ui/config.py

@@ -77,8 +77,6 @@ def validate(self):
 
 
 class Config(BaseModel):
-    security: SecurityComponent
-    service: ServiceComponent
     additional_components: Dict[str, Component] = {}
 
     _components: List[Component] = PrivateAttr(default_factory=list)
@@ -99,6 +97,11 @@ def context(self) -> Iterator[ConfigContext]:
         del self._ctx
 
 
+class AppConfig(Config):
+    security: SecurityComponent
+    service: ServiceComponent
+
+
 TConfig = TypeVar("TConfig", bound=Config)
 
 
@@ -127,6 +130,15 @@ def load_config(config_type: Type[TConfig], box: dict) -> TConfig:
     return config_type(additional_components=components, **named_components)
 
 
+def load_config_from_file(cls: Type[TConfig], path: str, envvar_prefix: str = "EVIDENTLY") -> TConfig:
+    dc = dynaconf.Dynaconf(
+        envvar_prefix=envvar_prefix,
+    )
+    dc.configure(settings_module=path)
+    config = load_config(cls, dc)
+    return config
+
+
 settings = dynaconf.Dynaconf(
     envvar_prefix="EVIDENTLY",
 )

src/evidently/ui/local_service.py

@@ -12,7 +12,7 @@
 from evidently.ui.components.storage import LocalStorageComponent
 from evidently.ui.components.storage import StorageComponent
 from evidently.ui.components.telemetry import TelemetryComponent
-from evidently.ui.config import Config
+from evidently.ui.config import AppConfig
 from evidently.ui.errors import EvidentlyServiceError
 
 
@@ -36,7 +36,7 @@ def apply(self, ctx: ComponentContext, builder: AppBuilder):
         builder.kwargs["debug"] = self.debug
 
 
-class LocalConfig(Config):
+class LocalConfig(AppConfig):
     security: SecurityComponent = NoSecurityComponent()
     service: ServiceComponent = LocalServiceComponent()
     storage: StorageComponent = LocalStorageComponent()

src/evidently/ui/storage/local/base.py

@@ -20,6 +20,7 @@
 from evidently.test_suite import TestSuite
 from evidently.tests.base_test import Test
 from evidently.tests.base_test import TestStatus
+from evidently.ui.base import BlobMetadata
 from evidently.ui.base import BlobStorage
 from evidently.ui.base import DataStorage
 from evidently.ui.base import MetadataStorage
@@ -84,6 +85,9 @@ def rmtree(self, path: str):
     def invalidate_cache(self, path):
         self.fs.invalidate_cache(posixpath.join(self.path, path))
 
+    def size(self, path):
+        return self.fs.size(posixpath.join(self.path, path))
+
 
 class FSSpecBlobStorage(BlobStorage):
     base_path: str
@@ -104,15 +108,18 @@ def get_snapshot_blob_id(self, project_id: ProjectID, snapshot: Snapshot) -> Blo
         return posixpath.join(str(project_id), SNAPSHOTS, str(snapshot.id)) + ".json"
 
     @contextlib.contextmanager
-    def open_blob(self, path: str):
-        with self.location.open(path) as f:
+    def open_blob(self, blob_id: str):
+        with self.location.open(blob_id) as f:
             yield f
 
-    def put_blob(self, path: str, obj) -> str:
-        self.location.makedirs(posixpath.dirname(path))
-        with self.location.open(path, "w") as f:
+    def put_blob(self, blob_id: BlobID, obj) -> BlobID:
+        self.location.makedirs(posixpath.dirname(blob_id))
+        with self.location.open(blob_id, "w") as f:
             f.write(obj)
-        return path
+        return blob_id
+
+    def get_blob_metadata(self, blob_id: BlobID) -> BlobMetadata:
+        return BlobMetadata(id=blob_id, size=self.location.size(blob_id))
 
 
 def load_project(location: FSLocation, path: str) -> Optional[Project]:
@@ -169,7 +176,9 @@ def reload_snapshot(self, project: Project, snapshot_id: SnapshotID, skip_errors
             snapshot_path = posixpath.join(str(project.id), SNAPSHOTS, str(snapshot_id) + ".json")
             with self.location.open(snapshot_path) as f:
                 suite = parse_obj_as(Snapshot, json.load(f))
-            snapshot = SnapshotMetadata.from_snapshot(suite, snapshot_path).bind(project)
+            snapshot = SnapshotMetadata.from_snapshot(
+                suite, BlobMetadata(id=snapshot_path, size=self.location.size(snapshot_path))
+            ).bind(project)
             self.snapshots[project.id][snapshot_id] = snapshot
             self.snapshot_data[project.id][snapshot_id] = suite
         except ValidationError as e:
@@ -220,11 +229,11 @@ def list_projects(self, project_ids: Optional[Set[ProjectID]]) -> List[Project]:
         projects.sort(key=lambda x: x.created_at or default_date, reverse=True)
         return projects
 
-    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob_id: str):
+    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob: BlobMetadata):
         project = self.get_project(project_id)
         if project is None:
             raise ProjectNotFound()
-        self.state.snapshots[project_id][snapshot.id] = SnapshotMetadata.from_snapshot(snapshot, blob_id).bind(project)
+        self.state.snapshots[project_id][snapshot.id] = SnapshotMetadata.from_snapshot(snapshot, blob).bind(project)
         self.state.snapshot_data[project_id][snapshot.id] = snapshot
 
     def delete_snapshot(self, project_id: ProjectID, snapshot_id: SnapshotID):

src/evidently/ui/workspace/remote.py

@@ -18,6 +18,7 @@
 from evidently._pydantic_compat import parse_obj_as
 from evidently.suite.base_suite import Snapshot
 from evidently.ui.api.service import EVIDENTLY_APPLICATION_NAME
+from evidently.ui.base import BlobMetadata
 from evidently.ui.base import BlobStorage
 from evidently.ui.base import DataStorage
 from evidently.ui.base import MetadataStorage
@@ -146,7 +147,7 @@ def delete_project(self, project_id: ProjectID):
     def list_projects(self, project_ids: Optional[Set[ProjectID]]) -> List[Project]:
         return self._request("/api/projects", "GET", response_model=List[Project])
 
-    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob_id: str):
+    def add_snapshot(self, project_id: ProjectID, snapshot: Snapshot, blob: "BlobMetadata"):
         return self._request(f"/api/projects/{project_id}/snapshots", "POST", body=snapshot.dict())
 
     def delete_snapshot(self, project_id: ProjectID, snapshot_id: SnapshotID):
@@ -182,7 +183,10 @@ def put_blob(self, path: str, obj):
         pass
 
     def get_snapshot_blob_id(self, project_id: ProjectID, snapshot: Snapshot) -> BlobID:
-        pass
+        return ""
+
+    def get_blob_metadata(self, blob_id: BlobID) -> BlobMetadata:
+        return BlobMetadata(id=blob_id, size=0)
 
 
 class NoopDataStorage(DataStorage):