src/rag/__init__.py

"""
RAG (Retrieval-Augmented Generation) Subsystem for SPARKNET

Provides:
- Vector store interface with ChromaDB implementation
- Embedding adapters (Ollama, OpenAI)
- Document indexing with metadata
- Grounded retrieval with evidence
- Answer generation with citations
"""

from .store import (
    VectorStoreConfig,
    VectorStore,
    VectorSearchResult,
    ChromaVectorStore,
    get_vector_store,
)

from .embeddings import (
    EmbeddingConfig,
    EmbeddingAdapter,
    OllamaEmbedding,
    get_embedding_adapter,
)

from .indexer import (
    IndexerConfig,
    IndexingResult,
    DocumentIndexer,
    get_document_indexer,
)

from .retriever import (
    RetrieverConfig,
    RetrievedChunk,
    DocumentRetriever,
    get_document_retriever,
)

from .generator import (
    GeneratorConfig,
    GeneratedAnswer,
    Citation,
    GroundedGenerator,
    get_grounded_generator,
)

from .docint_bridge import (
    DocIntIndexer,
    DocIntRetriever,
    get_docint_indexer,
    get_docint_retriever,
    reset_docint_components,
)

__all__ = [
    # Store
    "VectorStoreConfig",
    "VectorStore",
    "VectorSearchResult",
    "ChromaVectorStore",
    "get_vector_store",
    # Embeddings
    "EmbeddingConfig",
    "EmbeddingAdapter",
    "OllamaEmbedding",
    "get_embedding_adapter",
    # Indexer
    "IndexerConfig",
    "IndexingResult",
    "DocumentIndexer",
    "get_document_indexer",
    # Retriever
    "RetrieverConfig",
    "RetrievedChunk",
    "DocumentRetriever",
    "get_document_retriever",
    # Generator
    "GeneratorConfig",
    "GeneratedAnswer",
    "Citation",
    "GroundedGenerator",
    "get_grounded_generator",
    # Document Intelligence Bridge
    "DocIntIndexer",
    "DocIntRetriever",
    "get_docint_indexer",
    "get_docint_retriever",
    "reset_docint_components",
]