nirs4all.pipeline.storage.artifacts.artifact_registry module

Artifact Registry V3 - Central registry for artifact management.

This module provides the ArtifactRegistry class which serves as the central hub for artifact operations in the V3 artifacts system:

• Chain-based artifact identification for complete execution path tracking

• Content-addressed storage with global deduplication

• Dependency graph tracking for stacking/transfer

• Cleanup utilities for orphan detection and deletion

V3 Key Changes: - Uses OperatorChain for artifact identification instead of V2 ID format - Chain hash-based artifact IDs for deterministic identification - Unified handling of branching, multi-source, stacking, and bundles

The registry works with centralized storage at workspace/binaries/<dataset>/.

class nirs4all.pipeline.storage.artifacts.artifact_registry.DependencyGraph

Bases: object

Tracks artifact dependencies for stacking and transfer.

Maintains a directed graph where edges represent “depends on” relationships. Supports transitive dependency resolution with cycle detection.

add_dependencies(artifact_id: str, depends_on: list[str]) → None

Add multiple dependencies at once.

Parameters:

• artifact_id – The dependent artifact

• depends_on – List of artifacts being depended upon

add_dependency(artifact_id: str, depends_on: str) → None

Add a dependency relationship.

Parameters:

• artifact_id – The dependent artifact

• depends_on – The artifact being depended upon

clear() → None

Clear all dependencies.

get_dependencies(artifact_id: str) → list[str]

Get direct dependencies of an artifact.

Parameters:

artifact_id – Artifact to query

Returns:

List of artifact IDs this artifact depends on

get_dependents(artifact_id: str) → list[str]

Get artifacts that directly depend on this artifact.

Parameters:

artifact_id – Artifact to query

Returns:

List of artifact IDs that depend on this artifact

remove_artifact(artifact_id: str) → None

Remove an artifact and its edges from the graph.

Parameters:

artifact_id – Artifact to remove

resolve_dependencies(artifact_id: str, max_depth: int = 100) → list[str]

Get all transitive dependencies (topologically sorted).

Returns dependencies in order suitable for loading - dependencies before dependents.

Parameters:

• artifact_id – Starting artifact

• max_depth – Maximum recursion depth (prevents cycles)

Returns:

List of all dependencies in topological order

Raises:

ValueError – If cycle detected or max depth exceeded