src.dackar.RCA.orchestrators.artifact_store
===========================================

.. py:module:: src.dackar.RCA.orchestrators.artifact_store

.. autoapi-nested-parse::

   artifact_store — Concrete ArtifactStore and SchemaValidator implementations.

   Extracted from rca_reasoning_orchestrator.py.  The parent module re-exports
   both classes for backward-compatible imports.



Attributes
----------

.. autoapisummary::

   src.dackar.RCA.orchestrators.artifact_store.JsonDict


Classes
-------

.. autoapisummary::

   src.dackar.RCA.orchestrators.artifact_store.NoOpSchemaValidator
   src.dackar.RCA.orchestrators.artifact_store.FileArtifactStore


Module Contents
---------------

.. py:data:: JsonDict

.. py:class:: NoOpSchemaValidator

   Permissive :class:`SchemaValidator` that only enforces JSON-object shape.

   Implements all three validator styles but performs no schema checks — useful
   as a default when schema validation is not wired. Every artifact is reported
   as valid provided it is a JSON object.


   .. py:method:: validate(artifact_name, payload)

      Legacy style: raise ``TypeError`` unless *payload* is a JSON object.



   .. py:method:: validate_artifact(artifact_name, payload)

      Per-artifact style: shape-check *payload* and return an ``ok`` report.



   .. py:method:: validate_run_bundle(**kwargs)

      Bundle style: shape-check each non-None keyword artifact and return ``ok``.



.. py:class:: FileArtifactStore(root_dir)

   :class:`ArtifactStore` that persists artifacts as JSON files on disk.

   Each artifact is written atomically to ``<root_dir>/<run_id>/<name>.json``
   so a reader never observes a partial write.


   .. py:attribute:: root_dir


   .. py:method:: save(run_id, artifact_name, payload)

      Persist a single-object artifact and return its file path.



   .. py:method:: save_list(run_id, artifact_name, payload)

      Persist a list-valued artifact and return its file path.



   .. py:method:: load(run_id, artifact_name)

      Load and return a previously saved artifact, or None if absent.



   .. py:method:: is_run_complete(run_id)

      Return True only if run_status.json exists and run_complete is True.

      External callers (notebooks, replay scripts) should check this before
      loading or re-validating artifacts from a run directory.  A run that
      crashed mid-pipeline will have run_complete=False (or no run_status.json
      at all), and its artifacts should not be treated as authoritative.



   .. py:method:: _write_atomic(run_id, artifact_name, payload)

      Write payload to <run_dir>/<artifact_name>.json via temp-file + rename.

      The rename is atomic on POSIX and near-atomic on Windows (py3.3+).
      A reader can never observe a partial write.