src.dackar.RCA.storage.lc_retriever_processed
=============================================

.. py:module:: src.dackar.RCA.storage.lc_retriever_processed


Attributes
----------

.. autoapisummary::

   src.dackar.RCA.storage.lc_retriever_processed.LOGGER
   src.dackar.RCA.storage.lc_retriever_processed._WARNED_NON_RRF_FUSION


Classes
-------

.. autoapisummary::

   src.dackar.RCA.storage.lc_retriever_processed.RetrieveResult
   src.dackar.RCA.storage.lc_retriever_processed.ContextPack
   src.dackar.RCA.storage.lc_retriever_processed.LCProcessedRetriever


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

.. py:data:: LOGGER

.. py:data:: _WARNED_NON_RRF_FUSION
   :value: False


.. py:class:: RetrieveResult

   .. py:attribute:: record_id
      :type:  str


   .. py:attribute:: snippet
      :type:  str


   .. py:attribute:: metadata
      :type:  Dict[str, Any]


   .. py:attribute:: per_view
      :type:  Dict[str, Any]


   .. py:attribute:: score
      :type:  Optional[float]


   .. py:attribute:: record
      :type:  Optional[Dict[str, Any]]


.. py:class:: ContextPack

   .. py:attribute:: query_texts
      :type:  Dict[str, str]


   .. py:attribute:: results
      :type:  List[RetrieveResult]


.. py:class:: LCProcessedRetriever(manager, doc_store)

   Light adaptation of the earlier retrieval layer for processed_text_record.

   Differences from the original lc_retriever.py:
   - Uses ChromaRecordStore collections keyed by document type.
   - Treats record_id as the vector/document identity.
   - Hydrates canonical processed_text_record objects instead of raw mdParser chunks.


   .. py:attribute:: manager


   .. py:attribute:: store


   .. py:method:: query_doc_types(*, doc_types, query_text, top_k_per_doc_type = 8, k_final = 10, filter_meta = None, fusion = 'rrf', view_weights = None, hybrid_weight = 0.5, snippet_preference = 'raw_text')


   .. py:method:: _fuse(per_view_hits, *, fusion, k_final, view_weights)

      Fuse per-doc-type hit lists into a single ranked list.

      Only Reciprocal Rank Fusion (RRF) is valid at this cross-doc-type layer. Each incoming
      hit's ``score`` is the *fused RRF score* produced by ``ChromaRecordStore.query_doc_type``
      (higher = better), not a raw vector distance. ``weighted_distance_inversion`` assumes a
      raw non-negative distance (lower = better), so applying it here would rank matches
      worst-first. Any non-``"rrf"`` ``fusion`` value is therefore mapped to RRF with a
      one-time warning; ``weighted_distance_inversion`` remains available only for the
      per-view (raw-distance) layer inside the store.