dkist-processing-common 11.8.1rc2__tar.gz → 11.9.1__tar.gz

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/CHANGELOG.rst

@@ -1,3 +1,33 @@
+v11.9.1 (2025-12-08)
+====================
+
+Features
+--------
+
+- Store quality data in object store. (`#276 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/276>`__)
+
+
+v11.9.0 (2025-12-03)
+====================
+
+Misc
+----
+
+- Upgrade to the globus 4.x SDK. (`#274 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/274>`__)
+- Use a pool of direction aware globus accounts for transfers to mitigate contention between simultaneous inbound (TransferL0Data) and/or outbound (TransferL1Data) transfers. (`#274 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/274>`__)
+- Integrate dkist-processing-core 6.0.1 which adds additional attributes to metrics and tracing to facilitate discoverability and analysis. (`#275 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/275>`__)
+
+
+v11.8.1 (2025-12-02)
+====================
+
+Misc
+----
+
+- Bump minimum version of `pydantic` to 2.7.2 to avoid install failures for python versions >= 3.12. (`#271 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/271>`__)
+- Move `solar-wavelength-calibration` dep to 2.0.0 and make use of new helper properties in that release. (`#271 <https://bitbucket.org/dkistdc/dkist-processing-common/pull-requests/271>`__)
+
+
 v11.8.0 (2025-11-03)
 ====================
 

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dkist-processing-common
-Version: 11.8.1rc2
+Version: 11.9.1
 Summary: Common task classes used by the DKIST science data processing pipelines
 Author-email: NSO / AURA <dkistdc@nso.edu>
 License: BSD-3-Clause
@@ -17,12 +17,12 @@ Requires-Dist: asdf<4.0.0,>=3.5.0
 Requires-Dist: astropy>=7.0.0
 Requires-Dist: dkist-fits-specifications<5.0,>=4.0.0
 Requires-Dist: dkist-header-validator<6.0,>=5.0.0
-Requires-Dist: dkist-processing-core==6.0.0
+Requires-Dist: dkist-processing-core==6.0.1
 Requires-Dist: dkist-processing-pac<4.0,>=3.1
 Requires-Dist: dkist-service-configuration<5.0,>=4.1.7
 Requires-Dist: dkist-spectral-lines<4.0,>=3.0.0
-Requires-Dist: solar-wavelength-calibration<3.0,>=2.0.0rc3
-Requires-Dist: globus-sdk<4.0.0,>=3.12.0
+Requires-Dist: solar-wavelength-calibration<3.0,>=2.0.0
+Requires-Dist: globus-sdk<5.0.0,>=4.0.0
 Requires-Dist: gqlclient[pydantic]==1.2.3
 Requires-Dist: sqids==0.5.1
 Requires-Dist: matplotlib>=3.4
@@ -57,9 +57,9 @@ Requires-Dist: pytest; extra == "docs"
 Requires-Dist: towncrier<22.12.0; extra == "docs"
 Requires-Dist: dkist-sphinx-theme; extra == "docs"
 Provides-Extra: inventory
-Requires-Dist: dkist-inventory<2.0,>=1.11.1; extra == "inventory"
+Requires-Dist: dkist-inventory<2.0,>=1.11.2; extra == "inventory"
 Provides-Extra: asdf
-Requires-Dist: dkist-inventory[asdf]<2.0,>=1.11.1; extra == "asdf"
+Requires-Dist: dkist-inventory[asdf]<2.0,>=1.11.2; extra == "asdf"
 Provides-Extra: quality
 Requires-Dist: dkist-quality<3.0,>=2.0.0; extra == "quality"
 
@@ -118,6 +118,10 @@ Environment Variables
      - annotation=str required=False default='unknown-service-version' alias_priority=2 validation_alias='DKIST_SERVICE_VERSION' description='Service version for OpenTelemetry'
    * - NOMAD_ALLOC_ID
      - annotation=str required=False default='unknown-allocation-id' alias_priority=2 validation_alias='NOMAD_ALLOC_ID' description='Nomad allocation ID for OpenTelemetry'
+   * - NOMAD_ALLOC_NAME
+     - annotation=str required=False default='unknown-allocation-name' alias='NOMAD_ALLOC_NAME' alias_priority=2 description='Allocation name for the deployed container the task is running on.'
+   * - NOMAD_GROUP_NAME
+     - annotation=str required=False default='unknown-allocation-group' alias='NOMAD_GROUP_NAME' alias_priority=2 description='Allocation group for the deployed container the task is running on'
    * - OTEL_EXPORTER_OTLP_TRACES_INSECURE
      - annotation=bool required=False default=True description='Use insecure connection for OTLP traces'
    * - OTEL_EXPORTER_OTLP_METRICS_INSECURE
@@ -158,12 +162,12 @@ Environment Variables
      - annotation=Union[dict, NoneType] required=False default=None description='S3 upload configuration for the object store.'
    * - S3_DOWNLOAD_CONFIG
      - annotation=Union[dict, NoneType] required=False default=None description='S3 download configuration for the object store.'
-   * - GLOBUS_TRANSPORT_PARAMS
-     - annotation=dict required=False default_factory=dict description='Globus transfer parameters.'
-   * - GLOBUS_CLIENT_ID
-     - annotation=Union[str, NoneType] required=False default=None description='Globus client ID for inbound/outbound transfers.'
-   * - GLOBUS_CLIENT_SECRET
-     - annotation=Union[str, NoneType] required=False default=None description='Globus client secret for inbound/outbound transfers.'
+   * - GLOBUS_MAX_RETRIES
+     - annotation=int required=False default=5 description='Max retries for transient errors on calls to the globus api.'
+   * - GLOBUS_INBOUND_CLIENT_CREDENTIALS
+     - annotation=list[GlobusClientCredential] required=False default_factory=list description='Globus client credentials for inbound transfers.' examples=[[{'client_id': 'id1', 'client_secret': 'secret1'}, {'client_id': 'id2', 'client_secret': 'secret2'}]]
+   * - GLOBUS_OUTBOUND_CLIENT_CREDENTIALS
+     - annotation=list[GlobusClientCredential] required=False default_factory=list description='Globus client credentials for outbound transfers.' examples=[[{'client_id': 'id3', 'client_secret': 'secret3'}, {'client_id': 'id4', 'client_secret': 'secret4'}]]
    * - OBJECT_STORE_ENDPOINT
      - annotation=Union[str, NoneType] required=False default=None description='Object store Globus Endpoint ID.'
    * - SCRATCH_ENDPOINT

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/README.rst

@@ -53,6 +53,10 @@ Environment Variables
      - annotation=str required=False default='unknown-service-version' alias_priority=2 validation_alias='DKIST_SERVICE_VERSION' description='Service version for OpenTelemetry'
    * - NOMAD_ALLOC_ID
      - annotation=str required=False default='unknown-allocation-id' alias_priority=2 validation_alias='NOMAD_ALLOC_ID' description='Nomad allocation ID for OpenTelemetry'
+   * - NOMAD_ALLOC_NAME
+     - annotation=str required=False default='unknown-allocation-name' alias='NOMAD_ALLOC_NAME' alias_priority=2 description='Allocation name for the deployed container the task is running on.'
+   * - NOMAD_GROUP_NAME
+     - annotation=str required=False default='unknown-allocation-group' alias='NOMAD_GROUP_NAME' alias_priority=2 description='Allocation group for the deployed container the task is running on'
    * - OTEL_EXPORTER_OTLP_TRACES_INSECURE
      - annotation=bool required=False default=True description='Use insecure connection for OTLP traces'
    * - OTEL_EXPORTER_OTLP_METRICS_INSECURE
@@ -93,12 +97,12 @@ Environment Variables
      - annotation=Union[dict, NoneType] required=False default=None description='S3 upload configuration for the object store.'
    * - S3_DOWNLOAD_CONFIG
      - annotation=Union[dict, NoneType] required=False default=None description='S3 download configuration for the object store.'
-   * - GLOBUS_TRANSPORT_PARAMS
-     - annotation=dict required=False default_factory=dict description='Globus transfer parameters.'
-   * - GLOBUS_CLIENT_ID
-     - annotation=Union[str, NoneType] required=False default=None description='Globus client ID for inbound/outbound transfers.'
-   * - GLOBUS_CLIENT_SECRET
-     - annotation=Union[str, NoneType] required=False default=None description='Globus client secret for inbound/outbound transfers.'
+   * - GLOBUS_MAX_RETRIES
+     - annotation=int required=False default=5 description='Max retries for transient errors on calls to the globus api.'
+   * - GLOBUS_INBOUND_CLIENT_CREDENTIALS
+     - annotation=list[GlobusClientCredential] required=False default_factory=list description='Globus client credentials for inbound transfers.' examples=[[{'client_id': 'id1', 'client_secret': 'secret1'}, {'client_id': 'id2', 'client_secret': 'secret2'}]]
+   * - GLOBUS_OUTBOUND_CLIENT_CREDENTIALS
+     - annotation=list[GlobusClientCredential] required=False default_factory=list description='Globus client credentials for outbound transfers.' examples=[[{'client_id': 'id3', 'client_secret': 'secret3'}, {'client_id': 'id4', 'client_secret': 'secret4'}]]
    * - OBJECT_STORE_ENDPOINT
      - annotation=Union[str, NoneType] required=False default=None description='Object store Globus Endpoint ID.'
    * - SCRATCH_ENDPOINT

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/config.py

@@ -3,12 +3,20 @@
 from dkist_processing_core.config import DKISTProcessingCoreConfiguration
 from dkist_service_configuration.settings import DEFAULT_MESH_SERVICE
 from dkist_service_configuration.settings import MeshService
+from pydantic import BaseModel
 from pydantic import Field
 from talus import ConnectionRetryerFactory
 from talus import ConsumerConnectionParameterFactory
 from talus import ProducerConnectionParameterFactory
 
 
+class GlobusClientCredential(BaseModel):
+    """Globus client credential."""
+
+    client_id: str = Field(..., description="Globus client ID for transfers.")
+    client_secret: str = Field(..., description="Globus client secret for transfers.")
+
+
 class DKISTProcessingCommonConfiguration(DKISTProcessingCoreConfiguration):
     """Common configurations."""
 
@@ -40,14 +48,28 @@ class DKISTProcessingCommonConfiguration(DKISTProcessingCoreConfiguration):
         default=None, description="S3 download configuration for the object store."
     )
     # globus
-    globus_transport_params: dict = Field(
-        default_factory=dict, description="Globus transfer parameters."
+    globus_max_retries: int = Field(
+        default=5, description="Max retries for transient errors on calls to the globus api."
     )
-    globus_client_id: str | None = Field(
-        default=None, description="Globus client ID for inbound/outbound transfers."
+    globus_inbound_client_credentials: list[GlobusClientCredential] = Field(
+        default_factory=list,
+        description="Globus client credentials for inbound transfers.",
+        examples=[
+            [
+                {"client_id": "id1", "client_secret": "secret1"},
+                {"client_id": "id2", "client_secret": "secret2"},
+            ],
+        ],
     )
-    globus_client_secret: str | None = Field(
-        default=None, description="Globus client secret for inbound/outbound transfers."
+    globus_outbound_client_credentials: list[GlobusClientCredential] = Field(
+        default_factory=list,
+        description="Globus client credentials for outbound transfers.",
+        examples=[
+            [
+                {"client_id": "id3", "client_secret": "secret3"},
+                {"client_id": "id4", "client_secret": "secret4"},
+            ],
+        ],
     )
     object_store_endpoint: str | None = Field(
         default=None, description="Object store Globus Endpoint ID."

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/models/graphql.py

@@ -176,36 +176,3 @@ class RecipeRunProvenanceMutation(GraphqlBaseModel):
     libraryVersions: str
     workflowVersion: str
     codeVersion: str | None = None
-
-
-class QualityCreation(GraphqlBaseModel):
-    """Quality data creation record."""
-
-    datasetId: str
-    metricCode: str
-    facet: str | None = None
-    name: str | None = None
-    description: str | None = None
-    statement: str | None = None
-    # JSON array
-    warnings: str | None = None
-    # JSON objects
-    plotData: str | None = None
-    multiPlotData: str | None = None
-    tableData: str | None = None
-    histogramData: str | None = None
-    modmatData: str | None = None
-    raincloudData: str | None = None
-    efficiencyData: str | None = None
-
-
-class QualitiesRequest(GraphqlBaseModel):
-    """Query parameters for quality data."""
-
-    datasetId: str
-
-
-class QualityResponse(GraphqlBaseModel):
-    """Query Response for quality data."""
-
-    qualityId: int

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/l1_output_data.py

@@ -38,8 +38,9 @@ class L1OutputDataBase(OutputDataBase, ABC):
 
     @property
     def dataset_has_quality_data(self) -> bool:
-        """Return True if quality data has been persisted to the metadata-store."""
-        return self.metadata_store_quality_data_exists(dataset_id=self.constants.dataset_id)
+        """Return True if the dataset has quality data."""
+        path_count = self.count(tags=[Tag.output(), Tag.quality_data()])
+        return path_count > 0
 
     def rollback(self):
         """Warn that the metadata-store and the interservice bus retain the effect of this tasks execution.  Rolling back this task may not be achievable without other action."""
@@ -58,6 +59,9 @@ class TransferL1Data(TransferDataBase, GlobusMixin):
             # Movie needs to be transferred separately as the movie headers need to go with it
             self.transfer_movie()
 
+        with self.telemetry_span("Upload quality data"):
+            self.transfer_quality_data()
+
         with self.telemetry_span("Upload science frames"):
             self.transfer_output_frames()
 
@@ -101,6 +105,33 @@ class TransferL1Data(TransferDataBase, GlobusMixin):
             content_type="video/mp4",
         )
 
+    def transfer_quality_data(self):
+        """Transfer quality data to the object store."""
+        paths = list(self.read(tags=[Tag.output(), Tag.quality_data()]))
+        if len(paths) == 0:
+            logger.info(
+                f"No quality data found to upload for dataset. recipe_run_id={self.recipe_run_id}"
+            )
+            return
+
+        if count := len(paths) > 1:
+            # dataset inventory does not support multiple quality data object keys
+            raise RuntimeError(
+                f"Found multiple quality data files to upload.  Not supported."
+                f"{count=}, recipe_run_id={self.recipe_run_id}"
+            )
+
+        with self.telemetry_span(f"Uploading the trial quality data"):
+            path = paths[0]
+            logger.info(f"Uploading quality data: recipe_run_id={self.recipe_run_id}, {path=}")
+            quality_data_object_key = self.format_object_key(path)
+            self.object_store_upload_quality_data(
+                quality_data=path,
+                bucket=self.destination_bucket,
+                object_key=quality_data_object_key,
+                content_type="application/json",
+            )
+
 
 class AssembleQualityData(L1OutputDataBase, QualityMixin):
     """
@@ -128,7 +159,7 @@ class AssembleQualityData(L1OutputDataBase, QualityMixin):
         ):
             self.write(
                 quality_data,
-                tags=Tag.quality_data(),
+                tags=[Tag.output(), Tag.quality_data()],
                 encoder=quality_data_encoder,
                 relative_path=f"{self.constants.dataset_id}_quality_data.json",
             )
@@ -136,31 +167,18 @@ class AssembleQualityData(L1OutputDataBase, QualityMixin):
 
 class SubmitDatasetMetadata(L1OutputDataBase):
     """
-    Add quality data and receipt account to the metadata store.
+    Add receipt account to the metadata store.
 
-    Add the quality data to the Quality database.
     Add a Dataset Receipt Account record to Processing Support for use by the Dataset Catalog Locker.
-    Adds the number of files created during the calibration processing to the Processing Support table
+    Adds the number of files to be created during the calibration processing to the Processing Support table
     for use by the Dataset Catalog Locker.
     """
 
     def run(self) -> None:
         """Run method for this task."""
-        with self.telemetry_span(f"Storing quality data to metadata store"):
-            # each quality_data file is a list - this will combine the elements of multiple lists into a single list
-            quality_data = list(
-                chain.from_iterable(
-                    self.read(tags=Tag.quality_data(), decoder=quality_data_decoder)
-                )
-            )
-            self.metadata_store_add_quality_data(
-                dataset_id=self.constants.dataset_id, quality_data=quality_data
-            )
         with self.telemetry_span("Count Expected Outputs"):
             dataset_id = self.constants.dataset_id
             expected_object_count = self.count(tags=Tag.output())
-            if quality_data:
-                expected_object_count += 1
         logger.info(
             f"Adding Dataset Receipt Account: "
             f"{dataset_id=}, {expected_object_count=}, recipe_run_id={self.recipe_run_id}"
@@ -230,24 +248,12 @@ class PublishCatalogAndQualityMessages(L1OutputDataBase, InterserviceBusMixin):
         messages = [CatalogObjectMessage(body=body) for body in message_bodies]
         return messages
 
-    @property
-    def quality_report_message(self) -> CreateQualityReportMessage:
-        """Create the Quality Report Message."""
-        file_name = Path(f"{self.constants.dataset_id}_quality_report.pdf")
-        body = CreateQualityReportMessageBody(
-            bucket=self.destination_bucket,
-            objectName=self.format_object_key(file_name),
-            conversationId=str(self.recipe_run_id),
-            datasetId=self.constants.dataset_id,
-            incrementDatasetCatalogReceiptCount=True,
-        )
-        return CreateQualityReportMessage(body=body)
-
     def run(self) -> None:
         """Run method for this task."""
         with self.telemetry_span("Gather output data"):
             frames = self.read(tags=self.output_frame_tags)
             movies = self.read(tags=[Tag.output(), Tag.movie()])
+            quality_data = self.read(tags=[Tag.output(), Tag.quality_data()])
         with self.telemetry_span("Create message objects"):
             messages = []
             messages += self.frame_messages(paths=frames)
@@ -256,7 +262,7 @@ class PublishCatalogAndQualityMessages(L1OutputDataBase, InterserviceBusMixin):
             object_message_count = len(messages) - frame_message_count
             dataset_has_quality_data = self.dataset_has_quality_data
             if dataset_has_quality_data:
-                messages.append(self.quality_report_message)
+                messages += self.object_messages(paths=quality_data, object_type="QDATA")
         with self.telemetry_span(
             f"Publish messages: {frame_message_count = }, {object_message_count = }, {dataset_has_quality_data = }"
         ):

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/mixin/globus.py

@@ -9,6 +9,8 @@ from globus_sdk import ConfidentialAppAuthClient
 from globus_sdk import GlobusError
 from globus_sdk import TransferClient
 from globus_sdk import TransferData
+from globus_sdk.scopes import TransferScopes
+from globus_sdk.transport import RetryConfig
 
 from dkist_processing_common.config import common_configurations
 
@@ -31,27 +33,32 @@ class GlobusTransferItem:
 class GlobusMixin:
     """Mixin to add methods to a Task to support globus transfers."""
 
-    @property
-    def globus_transfer_client(self) -> TransferClient:
-        """Get the globus transfer client, creating it if it doesn't exist."""
-        if getattr(self, "_globus_transfer_client", False):
-            return self._globus_transfer_client
+    def globus_transfer_client_factory(self, transfer_data: TransferData) -> TransferClient:
+        """Create a globus transfer client based on the direction of transfer and round-robin the available application credentials."""
+        if (
+            transfer_data["source_endpoint"] == common_configurations.object_store_endpoint
+        ):  # inbound
+            client_credentials = common_configurations.globus_inbound_client_credentials
+        else:  # outbound
+            client_credentials = common_configurations.globus_outbound_client_credentials
+
+        # Round-robin the client credentials based on the recipe run id
+        index = self.recipe_run_id % len(client_credentials)
+        selected_credential = client_credentials[index]
+
         confidential_client = ConfidentialAppAuthClient(
-            client_id=common_configurations.globus_client_id,
-            client_secret=common_configurations.globus_client_secret,
-            transport_params=common_configurations.globus_transport_params,
+            client_id=selected_credential.client_id,
+            client_secret=selected_credential.client_secret,
         )
-        authorizer = ClientCredentialsAuthorizer(
-            confidential_client, scopes="urn:globus:auth:scope:transfer.api.globus.org:all"
-        )
-        self._globus_transfer_client = TransferClient(authorizer=authorizer)
-        return self._globus_transfer_client
+        authorizer = ClientCredentialsAuthorizer(confidential_client, scopes=TransferScopes)
+        retry_config = RetryConfig(max_retries=common_configurations.globus_max_retries)
+
+        return TransferClient(authorizer=authorizer, retry_config=retry_config)
 
     def globus_transfer_scratch_to_object_store(
         self,
         transfer_items: list[GlobusTransferItem],
         label: str = None,
-        sync_level: str = None,
         verify_checksum: bool = True,
     ) -> None:
         """Transfer data from scratch to the object store."""
@@ -60,7 +67,6 @@ class GlobusMixin:
             destination_endpoint=common_configurations.object_store_endpoint,
             transfer_items=transfer_items,
             label=label,
-            sync_level=sync_level,
             verify_checksum=verify_checksum,
         )
 
@@ -68,7 +74,6 @@ class GlobusMixin:
         self,
         transfer_items: list[GlobusTransferItem],
         label: str = None,
-        sync_level: str = None,
         verify_checksum: bool = True,
     ) -> None:
         """Transfer data from the object store to scratch."""
@@ -77,7 +82,6 @@ class GlobusMixin:
             destination_endpoint=common_configurations.scratch_endpoint,
             transfer_items=transfer_items,
             label=label,
-            sync_level=sync_level,
             verify_checksum=verify_checksum,
         )
 
@@ -87,7 +91,6 @@ class GlobusMixin:
         destination_endpoint: str,
         transfer_items: list[GlobusTransferItem],
         label: str = None,
-        sync_level: str = None,
         verify_checksum: bool = True,
     ) -> TransferData:
         """Format a globus TransferData instance."""
@@ -95,7 +98,6 @@ class GlobusMixin:
             source_endpoint=source_endpoint,
             destination_endpoint=destination_endpoint,
             label=label,
-            sync_level=sync_level,
             verify_checksum=verify_checksum,
         )
         for item in transfer_items:
@@ -112,7 +114,6 @@ class GlobusMixin:
         destination_endpoint: str,
         transfer_items: list[GlobusTransferItem],
         label: str = None,
-        sync_level: str = None,
         verify_checksum: bool = True,
     ) -> None:
         """Perform a transfer of data using globus."""
@@ -121,7 +122,6 @@ class GlobusMixin:
             destination_endpoint=destination_endpoint,
             transfer_items=transfer_items,
             label=label,
-            sync_level=sync_level,
             verify_checksum=verify_checksum,
         )
         self._blocking_globus_transfer(transfer_data=transfer_data)
@@ -131,24 +131,21 @@ class GlobusMixin:
         source_endpoint: str,
         destination_endpoint: str,
         label: str = None,
-        sync_level: str = None,
         verify_checksum: bool = True,
     ) -> TransferData:
         label = label or "Data Processing Transfer"
         return TransferData(
-            transfer_client=self.globus_transfer_client,
             source_endpoint=source_endpoint,
             destination_endpoint=destination_endpoint,
             label=label,
-            sync_level=sync_level,
             verify_checksum=verify_checksum,
         )
 
     def _blocking_globus_transfer(self, transfer_data: TransferData) -> None:
-        tc = self.globus_transfer_client
-        logger.info(f"Starting globus transfer: label={transfer_data.get('label')}")
+        tc = self.globus_transfer_client_factory(transfer_data=transfer_data)
         transfer_result = tc.submit_transfer(transfer_data)
         task_id = transfer_result["task_id"]
+        logger.info(f"Starting globus transfer: label={transfer_data.get('label')}, {task_id=}, ")
         polling_interval = 60
         while not tc.task_wait(
             task_id=task_id, timeout=polling_interval, polling_interval=polling_interval

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/mixin/metadata_store.py

@@ -1,6 +1,5 @@
 """Mixin for a WorkflowDataTaskBase subclass which implements Metadata Store data access functionality."""
 
-import json
 import logging
 from functools import cached_property
 from typing import Literal
@@ -8,15 +7,11 @@ from typing import Literal
 from pydantic import validate_call
 
 from dkist_processing_common._util.graphql import GraphQLClient
-from dkist_processing_common.codecs.quality import QualityDataEncoder
 from dkist_processing_common.config import common_configurations
 from dkist_processing_common.models.graphql import DatasetCatalogReceiptAccountMutation
 from dkist_processing_common.models.graphql import DatasetCatalogReceiptAccountResponse
 from dkist_processing_common.models.graphql import InputDatasetPartResponse
 from dkist_processing_common.models.graphql import InputDatasetRecipeRunResponse
-from dkist_processing_common.models.graphql import QualitiesRequest
-from dkist_processing_common.models.graphql import QualityCreation
-from dkist_processing_common.models.graphql import QualityResponse
 from dkist_processing_common.models.graphql import RecipeRunMutation
 from dkist_processing_common.models.graphql import RecipeRunMutationResponse
 from dkist_processing_common.models.graphql import RecipeRunProvenanceMutation
@@ -150,50 +145,6 @@ class MetadataStoreMixin:
             mutation_response_cls=RecipeRunProvenanceResponse,
         )
 
-    # QUALITY
-
-    def metadata_store_add_quality_data(self, dataset_id: str, quality_data: list[dict]):
-        """Add the quality data to the metadata-store."""
-        if self.metadata_store_quality_data_exists(dataset_id):
-            raise RuntimeError(f"Quality data already persisted for dataset {dataset_id!r}")
-        for metric in quality_data:
-            if (metric_code := metric.get("metric_code")) is None:
-                name = metric.get("name")
-                raise ValueError(f"No metric_code for {name!r} in dataset {dataset_id!r}")
-            params = QualityCreation(
-                datasetId=dataset_id,
-                metricCode=metric_code,
-                facet=metric.get("facet"),
-                name=metric.get("name"),
-                description=metric.get("description"),
-                statement=metric.get("statement"),
-                # JSON array
-                warnings=json.dumps(metric.get("warnings")),
-                # JSON objects
-                plotData=json.dumps(metric.get("plot_data"), cls=QualityDataEncoder),
-                multiPlotData=json.dumps(metric.get("multi_plot_data"), cls=QualityDataEncoder),
-                tableData=json.dumps(metric.get("table_data"), cls=QualityDataEncoder),
-                histogramData=json.dumps(metric.get("histogram_data"), cls=QualityDataEncoder),
-                modmatData=json.dumps(metric.get("modmat_data"), cls=QualityDataEncoder),
-                raincloudData=json.dumps(metric.get("raincloud_data"), cls=QualityDataEncoder),
-                efficiencyData=json.dumps(metric.get("efficiency_data"), cls=QualityDataEncoder),
-            )
-            self.metadata_store_client.execute_gql_mutation(
-                mutation_base="createQuality",
-                mutation_parameters=params,
-                mutation_response_cls=QualityResponse,
-            )
-
-    def metadata_store_quality_data_exists(self, dataset_id: str) -> bool:
-        """Return True if quality data exists in the metadata-store for the given dataset id."""
-        params = QualitiesRequest(datasetId=dataset_id)
-        response = self.metadata_store_client.execute_gql_query(
-            query_base="qualities",
-            query_response_cls=QualityResponse,
-            query_parameters=params,
-        )
-        return bool(response)
-
     # INPUT DATASET RECIPE RUN
 
     @cached_property

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/mixin/object_store.py

@@ -55,6 +55,27 @@ class ObjectStoreMixin:
             },
         )
 
+    def object_store_upload_quality_data(
+        self,
+        quality_data: Path | bytes,
+        bucket: str,
+        object_key: str,
+        content_type: str = "application/json",
+    ):
+        """Upload quality data to the object store."""
+        self.object_store_client.upload_object(
+            object_data=quality_data,
+            bucket=bucket,
+            object_key=object_key,
+            verify_checksum=True,
+            content_type=content_type,
+            metadata={
+                "groupname": "DATASET",
+                "groupid": self.constants.dataset_id,
+                "objecttype": "QDATA",
+            },
+        )
+
     def object_store_remove_folder_objects(self, bucket: str, path: Path | str) -> list[str]:
         """
         Remove folder objects (end with /) in the specified bucket and path.

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/trial_catalog.py

@@ -30,6 +30,7 @@ INVENTORY_EXTRA_INSTALLED = False
 try:
     from dkist_inventory.inventory import generate_asdf_filename
     from dkist_inventory.inventory import generate_inventory_from_frame_inventory
+    from dkist_inventory.inventory import generate_quality_report_filename
 
     INVENTORY_EXTRA_INSTALLED = True
 except ModuleNotFoundError:
@@ -224,6 +225,13 @@ class CreateTrialQualityReport(OutputDataBase):
                 f" but the required dependencies were not found."
             )
 
+        if not INVENTORY_EXTRA_INSTALLED:
+            raise ModuleNotFoundError(
+                f"{self.__class__.__name__} Task requires the dkist-inventory package "
+                f"(e.g. via an 'inventory' pip_extra on dkist_processing_core.Workflow().add_node())"
+                f" but the required dependencies were not found."
+            )
+
     def run(self) -> None:
         """Generate the quality report for the dataset."""
         self.create_trial_quality_report()
@@ -245,5 +253,7 @@ class CreateTrialQualityReport(OutputDataBase):
             self.write(
                 quality_report,
                 tags=[Tag.output(), Tag.quality_report()],
-                relative_path=f"{self.constants.dataset_id}_quality_report.pdf",
+                relative_path=generate_quality_report_filename(
+                    dataset_id=self.constants.dataset_id
+                ),
             )

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tasks/trial_output_data.py

@@ -100,7 +100,7 @@ class TransferTrialData(TransferDataBase, GlobusMixin):
         tag_list = []
         tag_list += [[Tag.output(), Tag.dataset_inventory()]]
         tag_list += [[Tag.output(), Tag.asdf()]]
-        tag_list += [[Tag.quality_data()]]  # quality data is not tagged as OUTPUT
+        tag_list += [[Tag.output(), Tag.quality_data()]]
         tag_list += [[Tag.output(), Tag.quality_report()]]
         tag_list += [[Tag.output(), Tag.movie()]]
         return tag_list

{dkist_processing_common-11.8.1rc2 → dkist_processing_common-11.9.1}/dkist_processing_common/tests/mock_metadata_store.py

@@ -135,10 +135,6 @@ class InputDatasetRecipeRunResponseMapping(ResponseMapping):
         return Unset
 
 
-class QualityResponseMapping(ResponseMapping):
-    pass  # TODO
-
-
 def make_default_recipe_run_status_response() -> RecipeRunStatusResponse:
     return RecipeRunStatusResponse(recipeRunStatusId=1)