zenml-nightly 0.70.0.dev20241125__py3-none-any.whl → 0.71.0.dev20241220__py3-none-any.whl

zenml/zen_stores/schemas/pipeline_run_schemas.py

@@ -15,7 +15,7 @@
 
 import json
 from datetime import datetime
-from typing import TYPE_CHECKING, Any, List, Optional
+from typing import TYPE_CHECKING, Any, Dict, List, Optional
 from uuid import UUID
 
 from pydantic import ConfigDict
@@ -34,6 +34,7 @@ from zenml.models import (
     PipelineRunResponseBody,
     PipelineRunResponseMetadata,
     PipelineRunUpdate,
+    RunMetadataEntry,
 )
 from zenml.models.v2.core.pipeline_run import PipelineRunResponseResources
 from zenml.zen_stores.schemas.base_schemas import NamedSchema
@@ -48,6 +49,7 @@ from zenml.zen_stores.schemas.schema_utils import build_foreign_key_field
 from zenml.zen_stores.schemas.stack_schemas import StackSchema
 from zenml.zen_stores.schemas.trigger_schemas import TriggerExecutionSchema
 from zenml.zen_stores.schemas.user_schemas import UserSchema
+from zenml.zen_stores.schemas.utils import RunMetadataInterface
 from zenml.zen_stores.schemas.workspace_schemas import WorkspaceSchema
 
 if TYPE_CHECKING:
@@ -56,13 +58,15 @@ if TYPE_CHECKING:
         ModelVersionPipelineRunSchema,
         ModelVersionSchema,
     )
-    from zenml.zen_stores.schemas.run_metadata_schemas import RunMetadataSchema
+    from zenml.zen_stores.schemas.run_metadata_schemas import (
+        RunMetadataResourceSchema,
+    )
     from zenml.zen_stores.schemas.service_schemas import ServiceSchema
     from zenml.zen_stores.schemas.step_run_schemas import StepRunSchema
     from zenml.zen_stores.schemas.tag_schemas import TagResourceSchema
 
 
-class PipelineRunSchema(NamedSchema, table=True):
+class PipelineRunSchema(NamedSchema, RunMetadataInterface, table=True):
     """SQL Model for pipeline runs."""
 
     __tablename__ = "pipeline_run"
@@ -136,12 +140,12 @@ class PipelineRunSchema(NamedSchema, table=True):
     )
     workspace: "WorkspaceSchema" = Relationship(back_populates="runs")
     user: Optional["UserSchema"] = Relationship(back_populates="runs")
-    run_metadata: List["RunMetadataSchema"] = Relationship(
-        back_populates="pipeline_run",
+    run_metadata_resources: List["RunMetadataResourceSchema"] = Relationship(
+        back_populates="pipeline_runs",
         sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.PIPELINE_RUN.value}', foreign(RunMetadataSchema.resource_id)==PipelineRunSchema.id)",
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.PIPELINE_RUN.value}', foreign(RunMetadataResourceSchema.resource_id)==PipelineRunSchema.id)",
             cascade="delete",
-            overlaps="run_metadata",
+            overlaps="run_metadata_resources",
         ),
     )
     logs: Optional["LogsSchema"] = Relationship(
@@ -249,6 +253,24 @@ class PipelineRunSchema(NamedSchema, table=True):
             model_version_id=request.model_version_id,
         )
 
+    def fetch_metadata_collection(self) -> Dict[str, List[RunMetadataEntry]]:
+        """Fetches all the metadata entries related to the pipeline run.
+
+        Returns:
+            a dictionary, where the key is the key of the metadata entry
+                and the values represent the list of entries with this key.
+        """
+        # Fetch the metadata related to this run
+        metadata_collection = super().fetch_metadata_collection()
+
+        # Fetch the metadata related to the steps of this run
+        for s in self.step_runs:
+            step_metadata = s.fetch_metadata_collection()
+            for k, v in step_metadata.items():
+                metadata_collection[f"{s.name}::{k}"] = v
+
+        return metadata_collection
+
     def to_model(
         self,
         include_metadata: bool = False,
@@ -275,15 +297,14 @@ class PipelineRunSchema(NamedSchema, table=True):
             else {}
         )
 
-        run_metadata = {
-            metadata_schema.key: json.loads(metadata_schema.value)
-            for metadata_schema in self.run_metadata
-        }
-
         if self.deployment is not None:
-            deployment = self.deployment.to_model()
+            deployment = self.deployment.to_model(include_metadata=True)
 
             config = deployment.pipeline_configuration
+            new_substitutions = config._get_full_substitutions(self.start_time)
+            config = config.model_copy(
+                update={"substitutions": new_substitutions}
+            )
             client_environment = deployment.client_environment
 
             stack = deployment.stack
@@ -323,9 +344,11 @@ class PipelineRunSchema(NamedSchema, table=True):
             build=build,
             schedule=schedule,
             code_reference=code_reference,
-            trigger_execution=self.trigger_execution.to_model()
-            if self.trigger_execution
-            else None,
+            trigger_execution=(
+                self.trigger_execution.to_model()
+                if self.trigger_execution
+                else None
+            ),
             created=self.created,
             updated=self.updated,
             deployment_id=self.deployment_id,
@@ -342,11 +365,21 @@ class PipelineRunSchema(NamedSchema, table=True):
             ):
                 is_templatable = True
 
-            steps = {step.name: step.to_model() for step in self.step_runs}
+            steps = {
+                step.name: step.to_model(include_metadata=True)
+                for step in self.step_runs
+            }
+
+            step_substitutions = {}
+            for step_name, step in steps.items():
+                step_substitutions[step_name] = step.config.substitutions
+                # We fetch the steps hydrated before, but want them unhydrated
+                # in the response -> We need to reset the metadata here
+                step.metadata = None
 
             metadata = PipelineRunResponseMetadata(
                 workspace=self.workspace.to_model(),
-                run_metadata=run_metadata,
+                run_metadata=self.fetch_metadata(),
                 config=config,
                 steps=steps,
                 start_time=self.start_time,
@@ -361,6 +394,7 @@ class PipelineRunSchema(NamedSchema, table=True):
                 if self.deployment
                 else None,
                 is_templatable=is_templatable,
+                step_substitutions=step_substitutions,
             )
 
         resources = None

zenml/zen_stores/schemas/pipeline_schemas.py

@@ -156,7 +156,12 @@ class PipelineSchema(NamedSchema, table=True):
 
         resources = None
         if include_resources:
+            latest_run_user = self.runs[-1].user if self.runs else None
+
             resources = PipelineResponseResources(
+                latest_run_user=latest_run_user.to_model()
+                if latest_run_user
+                else None,
                 tags=[t.tag.to_model() for t in self.tags],
             )
 

zenml/zen_stores/schemas/run_metadata_schemas.py

@@ -14,15 +14,16 @@
 """SQLModel implementation of pipeline run metadata tables."""
 
 from typing import TYPE_CHECKING, List, Optional
-from uuid import UUID
+from uuid import UUID, uuid4
 
 from sqlalchemy import TEXT, VARCHAR, Column
-from sqlmodel import Field, Relationship
+from sqlmodel import Field, Relationship, SQLModel
 
 from zenml.enums import MetadataResourceTypes
 from zenml.zen_stores.schemas.base_schemas import BaseSchema
 from zenml.zen_stores.schemas.component_schemas import StackComponentSchema
 from zenml.zen_stores.schemas.schema_utils import build_foreign_key_field
+from zenml.zen_stores.schemas.step_run_schemas import StepRunSchema
 from zenml.zen_stores.schemas.user_schemas import UserSchema
 from zenml.zen_stores.schemas.workspace_schemas import WorkspaceSchema
 
@@ -30,7 +31,6 @@ if TYPE_CHECKING:
     from zenml.zen_stores.schemas.artifact_schemas import ArtifactVersionSchema
     from zenml.zen_stores.schemas.model_schemas import ModelVersionSchema
     from zenml.zen_stores.schemas.pipeline_run_schemas import PipelineRunSchema
-    from zenml.zen_stores.schemas.step_run_schemas import StepRunSchema
 
 
 class RunMetadataSchema(BaseSchema, table=True):
@@ -38,35 +38,10 @@ class RunMetadataSchema(BaseSchema, table=True):
 
     __tablename__ = "run_metadata"
 
-    resource_id: UUID
-    resource_type: str = Field(sa_column=Column(VARCHAR(255), nullable=False))
-    pipeline_run: List["PipelineRunSchema"] = Relationship(
-        back_populates="run_metadata",
-        sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.PIPELINE_RUN.value}', foreign(RunMetadataSchema.resource_id)==PipelineRunSchema.id)",
-            overlaps="run_metadata,step_run,artifact_version,model_version",
-        ),
-    )
-    step_run: List["StepRunSchema"] = Relationship(
-        back_populates="run_metadata",
-        sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.STEP_RUN.value}', foreign(RunMetadataSchema.resource_id)==StepRunSchema.id)",
-            overlaps="run_metadata,pipeline_run,artifact_version,model_version",
-        ),
-    )
-    artifact_version: List["ArtifactVersionSchema"] = Relationship(
-        back_populates="run_metadata",
-        sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.ARTIFACT_VERSION.value}', foreign(RunMetadataSchema.resource_id)==ArtifactVersionSchema.id)",
-            overlaps="run_metadata,pipeline_run,step_run,model_version",
-        ),
-    )
-    model_version: List["ModelVersionSchema"] = Relationship(
+    # Relationship to link to resources
+    resources: List["RunMetadataResourceSchema"] = Relationship(
         back_populates="run_metadata",
-        sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.MODEL_VERSION.value}', foreign(RunMetadataSchema.resource_id)==ModelVersionSchema.id)",
-            overlaps="run_metadata,pipeline_run,step_run,artifact_version",
-        ),
+        sa_relationship_kwargs={"cascade": "delete"},
     )
     stack_component_id: Optional[UUID] = build_foreign_key_field(
         source=__tablename__,
@@ -103,3 +78,63 @@ class RunMetadataSchema(BaseSchema, table=True):
     key: str
     value: str = Field(sa_column=Column(TEXT, nullable=False))
     type: str
+
+    publisher_step_id: Optional[UUID] = build_foreign_key_field(
+        source=__tablename__,
+        target=StepRunSchema.__tablename__,
+        source_column="publisher_step_id",
+        target_column="id",
+        ondelete="SET NULL",
+        nullable=True,
+    )
+
+
+class RunMetadataResourceSchema(SQLModel, table=True):
+    """Table for linking resources to run metadata entries."""
+
+    __tablename__ = "run_metadata_resource"
+
+    id: UUID = Field(default_factory=uuid4, primary_key=True)
+    resource_id: UUID
+    resource_type: str = Field(sa_column=Column(VARCHAR(255), nullable=False))
+    run_metadata_id: UUID = build_foreign_key_field(
+        source=__tablename__,
+        target=RunMetadataSchema.__tablename__,
+        source_column="run_metadata_id",
+        target_column="id",
+        ondelete="CASCADE",
+        nullable=False,
+    )
+
+    # Relationship back to the base metadata table
+    run_metadata: RunMetadataSchema = Relationship(back_populates="resources")
+
+    # Relationship to link specific resource types
+    pipeline_runs: List["PipelineRunSchema"] = Relationship(
+        back_populates="run_metadata_resources",
+        sa_relationship_kwargs=dict(
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.PIPELINE_RUN.value}', foreign(RunMetadataResourceSchema.resource_id)==PipelineRunSchema.id)",
+            overlaps="run_metadata_resources,step_runs,artifact_versions,model_versions",
+        ),
+    )
+    step_runs: List["StepRunSchema"] = Relationship(
+        back_populates="run_metadata_resources",
+        sa_relationship_kwargs=dict(
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.STEP_RUN.value}', foreign(RunMetadataResourceSchema.resource_id)==StepRunSchema.id)",
+            overlaps="run_metadata_resources,pipeline_runs,artifact_versions,model_versions",
+        ),
+    )
+    artifact_versions: List["ArtifactVersionSchema"] = Relationship(
+        back_populates="run_metadata_resources",
+        sa_relationship_kwargs=dict(
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.ARTIFACT_VERSION.value}', foreign(RunMetadataResourceSchema.resource_id)==ArtifactVersionSchema.id)",
+            overlaps="run_metadata_resources,pipeline_runs,step_runs,model_versions",
+        ),
+    )
+    model_versions: List["ModelVersionSchema"] = Relationship(
+        back_populates="run_metadata_resources",
+        sa_relationship_kwargs=dict(
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.MODEL_VERSION.value}', foreign(RunMetadataResourceSchema.resource_id)==ModelVersionSchema.id)",
+            overlaps="run_metadata_resources,pipeline_runs,step_runs,artifact_versions",
+        ),
+    )

zenml/zen_stores/schemas/step_run_schemas.py

@@ -19,10 +19,11 @@ from typing import TYPE_CHECKING, Any, Dict, List, Optional
 from uuid import UUID
 
 from pydantic import ConfigDict
-from sqlalchemy import TEXT, Column, String
+from sqlalchemy import TEXT, Column, String, UniqueConstraint
 from sqlalchemy.dialects.mysql import MEDIUMTEXT
 from sqlmodel import Field, Relationship, SQLModel
 
+from zenml.config.pipeline_configurations import PipelineConfiguration
 from zenml.config.step_configurations import Step
 from zenml.constants import MEDIUMTEXT_MAX_LENGTH
 from zenml.enums import (
@@ -50,19 +51,29 @@ from zenml.zen_stores.schemas.pipeline_deployment_schemas import (
 from zenml.zen_stores.schemas.pipeline_run_schemas import PipelineRunSchema
 from zenml.zen_stores.schemas.schema_utils import build_foreign_key_field
 from zenml.zen_stores.schemas.user_schemas import UserSchema
+from zenml.zen_stores.schemas.utils import RunMetadataInterface
 from zenml.zen_stores.schemas.workspace_schemas import WorkspaceSchema
 
 if TYPE_CHECKING:
     from zenml.zen_stores.schemas.artifact_schemas import ArtifactVersionSchema
     from zenml.zen_stores.schemas.logs_schemas import LogsSchema
     from zenml.zen_stores.schemas.model_schemas import ModelVersionSchema
-    from zenml.zen_stores.schemas.run_metadata_schemas import RunMetadataSchema
+    from zenml.zen_stores.schemas.run_metadata_schemas import (
+        RunMetadataResourceSchema,
+    )
 
 
-class StepRunSchema(NamedSchema, table=True):
+class StepRunSchema(NamedSchema, RunMetadataInterface, table=True):
     """SQL Model for steps of pipeline runs."""
 
     __tablename__ = "step_run"
+    __table_args__ = (
+        UniqueConstraint(
+            "name",
+            "pipeline_run_id",
+            name="unique_step_name_for_pipeline_run",
+        ),
+    )
 
     # Fields
     start_time: Optional[datetime] = Field(nullable=True)
@@ -139,12 +150,12 @@ class StepRunSchema(NamedSchema, table=True):
     deployment: Optional["PipelineDeploymentSchema"] = Relationship(
         back_populates="step_runs"
     )
-    run_metadata: List["RunMetadataSchema"] = Relationship(
-        back_populates="step_run",
+    run_metadata_resources: List["RunMetadataResourceSchema"] = Relationship(
+        back_populates="step_runs",
         sa_relationship_kwargs=dict(
-            primaryjoin=f"and_(RunMetadataSchema.resource_type=='{MetadataResourceTypes.STEP_RUN.value}', foreign(RunMetadataSchema.resource_id)==StepRunSchema.id)",
+            primaryjoin=f"and_(RunMetadataResourceSchema.resource_type=='{MetadataResourceTypes.STEP_RUN.value}', foreign(RunMetadataResourceSchema.resource_id)==StepRunSchema.id)",
             cascade="delete",
-            overlaps="run_metadata",
+            overlaps="run_metadata_resources",
         ),
     )
     input_artifacts: List["StepRunInputArtifactSchema"] = Relationship(
@@ -163,9 +174,15 @@ class StepRunSchema(NamedSchema, table=True):
             "primaryjoin": "StepRunParentsSchema.child_id == StepRunSchema.id",
         },
     )
+    pipeline_run: "PipelineRunSchema" = Relationship(
+        back_populates="step_runs"
+    )
     model_version: "ModelVersionSchema" = Relationship(
         back_populates="step_runs",
     )
+    original_step_run: Optional["StepRunSchema"] = Relationship(
+        sa_relationship_kwargs={"remote_side": "StepRunSchema.id"}
+    )
 
     model_config = ConfigDict(protected_namespaces=())  # type: ignore[assignment]
 
@@ -218,11 +235,6 @@ class StepRunSchema(NamedSchema, table=True):
             RuntimeError: If the step run schema does not have a deployment_id
                 or a step_configuration.
         """
-        run_metadata = {
-            metadata_schema.key: json.loads(metadata_schema.value)
-            for metadata_schema in self.run_metadata
-        }
-
         input_artifacts = {
             artifact.name: StepRunInputResponse(
                 input_type=StepRunInputArtifactType(artifact.type),
@@ -248,6 +260,21 @@ class StepRunSchema(NamedSchema, table=True):
                 full_step_config = Step.model_validate(
                     step_configuration[self.name]
                 )
+                new_substitutions = (
+                    full_step_config.config._get_full_substitutions(
+                        PipelineConfiguration.model_validate_json(
+                            self.deployment.pipeline_configuration
+                        ),
+                        self.pipeline_run.start_time,
+                    )
+                )
+                full_step_config = full_step_config.model_copy(
+                    update={
+                        "config": full_step_config.config.model_copy(
+                            update={"substitutions": new_substitutions}
+                        )
+                    }
+                )
             elif not self.step_configuration:
                 raise ValueError(
                     f"Unable to load the configuration for step `{self.name}` from the"
@@ -294,7 +321,7 @@ class StepRunSchema(NamedSchema, table=True):
                 pipeline_run_id=self.pipeline_run_id,
                 original_step_run_id=self.original_step_run_id,
                 parent_step_ids=[p.parent_id for p in self.parents],
-                run_metadata=run_metadata,
+                run_metadata=self.fetch_metadata(),
             )
 
         resources = None

zenml/zen_stores/schemas/utils.py

@@ -13,11 +13,14 @@
 #  permissions and limitations under the License.
 """Utils for schemas."""
 
+import json
 import math
-from typing import List, Type, TypeVar
+from typing import Dict, List, Type, TypeVar
 
-from zenml.models.v2.base.base import BaseResponse
-from zenml.models.v2.base.page import Page
+from sqlmodel import Relationship
+
+from zenml.metadata.metadata_types import MetadataType
+from zenml.models import BaseResponse, Page, RunMetadataEntry
 from zenml.zen_stores.schemas.base_schemas import BaseSchema
 
 S = TypeVar("S", bound=BaseSchema)
@@ -67,3 +70,44 @@ def get_page_from_list(
         total=total,
         items=page_items,
     )
+
+
+class RunMetadataInterface:
+    """The interface for entities with run metadata."""
+
+    run_metadata_resources = Relationship()
+
+    def fetch_metadata_collection(self) -> Dict[str, List[RunMetadataEntry]]:
+        """Fetches all the metadata entries related to the artifact version.
+
+        Returns:
+            a dictionary, where the key is the key of the metadata entry
+                and the values represent the list of entries with this key.
+        """
+        metadata_collection: Dict[str, List[RunMetadataEntry]] = {}
+
+        # Fetch the metadata related to this step
+        for rm in self.run_metadata_resources:
+            if rm.run_metadata.key not in metadata_collection:
+                metadata_collection[rm.run_metadata.key] = []
+            metadata_collection[rm.run_metadata.key].append(
+                RunMetadataEntry(
+                    value=json.loads(rm.run_metadata.value),
+                    created=rm.run_metadata.created,
+                )
+            )
+
+        return metadata_collection
+
+    def fetch_metadata(self) -> Dict[str, MetadataType]:
+        """Fetches the latest metadata entry related to the artifact version.
+
+        Returns:
+            a dictionary, where the key is the key of the metadata entry
+                and the values represent the latest entry with this key.
+        """
+        metadata_collection = self.fetch_metadata_collection()
+        return {
+            k: sorted(v, key=lambda x: x.created, reverse=True)[0].value
+            for k, v in metadata_collection.items()
+        }