rag.core.config
===============

.. py:module:: rag.core.config

.. autoapi-nested-parse::

   Configuration for the papers RAG pipeline.



Attributes
----------

.. autoapisummary::

   rag.core.config.PAPERS_DIR_ENV
   rag.core.config.CHROMA_DIR_ENV
   rag.core.config.CHROMA_COLLECTION_ENV
   rag.core.config.EMBED_MODEL_ENV
   rag.core.config.OPENAI_API_KEY_ENV
   rag.core.config.OPENAI_MODEL_ENV
   rag.core.config.CHUNK_SIZE_ENV
   rag.core.config.CHUNK_OVERLAP_ENV
   rag.core.config.SIMILARITY_TOP_K_ENV
   rag.core.config.DEFAULT_COLLECTION_NAME
   rag.core.config.DEFAULT_EMBED_MODEL
   rag.core.config.DEFAULT_LLM_MODEL
   rag.core.config.DEFAULT_CHUNK_SIZE
   rag.core.config.DEFAULT_CHUNK_OVERLAP
   rag.core.config.DEFAULT_SIMILARITY_TOP_K


Classes
-------

.. autoapisummary::

   rag.core.config.RagConfig


Functions
---------

.. autoapisummary::

   rag.core.config.find_repo_root
   rag.core.config.load_repo_dotenv


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

.. py:data:: PAPERS_DIR_ENV
   :type:  Final
   :value: 'PAPERS_DIR'


.. py:data:: CHROMA_DIR_ENV
   :type:  Final
   :value: 'CHROMA_DIR'


.. py:data:: CHROMA_COLLECTION_ENV
   :type:  Final
   :value: 'CHROMA_COLLECTION'


.. py:data:: EMBED_MODEL_ENV
   :type:  Final
   :value: 'EMBED_MODEL'


.. py:data:: OPENAI_API_KEY_ENV
   :type:  Final
   :value: 'OPENAI_API_KEY'


.. py:data:: OPENAI_MODEL_ENV
   :type:  Final
   :value: 'OPENAI_MODEL'


.. py:data:: CHUNK_SIZE_ENV
   :type:  Final
   :value: 'CHUNK_SIZE'


.. py:data:: CHUNK_OVERLAP_ENV
   :type:  Final
   :value: 'CHUNK_OVERLAP'


.. py:data:: SIMILARITY_TOP_K_ENV
   :type:  Final
   :value: 'SIMILARITY_TOP_K'


.. py:data:: DEFAULT_COLLECTION_NAME
   :type:  Final
   :value: 'papers'


.. py:data:: DEFAULT_EMBED_MODEL
   :type:  Final
   :value: 'BAAI/bge-small-en-v1.5'


.. py:data:: DEFAULT_LLM_MODEL
   :type:  Final
   :value: 'gpt-4o-mini'


.. py:data:: DEFAULT_CHUNK_SIZE
   :type:  Final
   :value: 1024


.. py:data:: DEFAULT_CHUNK_OVERLAP
   :type:  Final
   :value: 128


.. py:data:: DEFAULT_SIMILARITY_TOP_K
   :type:  Final
   :value: 5


.. py:function:: find_repo_root(*, start: pathlib.Path | None = None) -> pathlib.Path

   Walk parents until the monorepo root (contains ``libs/rag-core``) is found.

   :param start: Directory to start from. Defaults to this file's location.
   :type start: Path or None

   :returns: Absolute path to the monorepo root.
   :rtype: Path

   :raises FileNotFoundError: If no ancestor contains ``libs/rag-core``.


.. py:function:: load_repo_dotenv() -> None

   Load ``.env`` from cwd and the monorepo root.

   Existing process environment wins (``override=False``). Streamlit sets
   cwd to the script directory, so a repo-root ``.env`` is missed by a
   bare ``load_dotenv()``.


.. py:class:: RagConfig

   Runtime settings for ingest + query.

   :ivar papers_dir: Directory of PDF papers to ingest.
   :vartype papers_dir: Path
   :ivar chroma_dir: Persistent Chroma directory (created if missing).
   :vartype chroma_dir: Path
   :ivar collection_name: Chroma collection name.
   :vartype collection_name: str
   :ivar embed_model_name: HuggingFace embedding model id (local).
   :vartype embed_model_name: str
   :ivar llm_model_name: OpenAI chat model id.
   :vartype llm_model_name: str
   :ivar chunk_size: SentenceSplitter chunk size.
   :vartype chunk_size: int
   :ivar chunk_overlap: SentenceSplitter overlap.
   :vartype chunk_overlap: int
   :ivar similarity_top_k: Retrieval top-k for the query engine.
   :vartype similarity_top_k: int
   :ivar strategy: In-process orchestration backend (LlamaIndex or LangChain).

   :vartype strategy: RagStrategy


   .. py:attribute:: papers_dir
      :type:  pathlib.Path


   .. py:attribute:: chroma_dir
      :type:  pathlib.Path


   .. py:attribute:: collection_name
      :type:  str
      :value: 'papers'



   .. py:attribute:: embed_model_name
      :type:  str
      :value: 'BAAI/bge-small-en-v1.5'



   .. py:attribute:: llm_model_name
      :type:  str
      :value: 'gpt-4o-mini'



   .. py:attribute:: chunk_size
      :type:  int
      :value: 1024



   .. py:attribute:: chunk_overlap
      :type:  int
      :value: 128



   .. py:attribute:: similarity_top_k
      :type:  int
      :value: 5



   .. py:attribute:: strategy
      :type:  rag.core.strategy.RagStrategy


   .. py:method:: from_env(*, repo_root: pathlib.Path | None = None, strategy: rag.core.strategy.RagStrategy | None = None) -> RagConfig
      :classmethod:


      Build config from environment variables with repo-relative defaults.

      :param repo_root: Monorepo root. Discovered automatically when omitted.
      :type repo_root: Path or None
      :param strategy: Orchestration backend. Defaults to ``RAG_STRATEGY`` or LlamaIndex.
      :type strategy: RagStrategy or None

      :returns: Resolved configuration.
      :rtype: RagConfig



   .. py:method:: ensure_dirs() -> None

      Create the Chroma persistence directory if needed.