RAGLibrary

class RAGLibrary:
    def __init__(
        self,
        # Component selection
        rag_method: str = "hybrid_graph",
        chunker: str = "sentence",
        embedder: str = "gemini",
        vector_store: str = "chroma",
        graph_store: str = "neo4j",
        llm: str = "gemini",
        extraction: str = "hybrid",
        ocr: str = "gemini_vision",
        page_index: str = "hybrid",
        # API credentials
        gemini_api_key: str | None = None,
        openai_api_key: str | None = None,
        anthropic_api_key: str | None = None,
        azure_openai_endpoint: str | None = None,
        azure_openai_key: str | None = None,
        aws_region: str = "us-east-1",
        neo4j_uri: str = "bolt://localhost:7687",
        neo4j_user: str = "neo4j",
        neo4j_password: str | None = None,
        cohere_api_key: str | None = None,
        ollama_base_url: str = "http://localhost:11434",
        # Advanced: full config override
        config: LibraryConfig | None = None,
    )

register_loader(extension: str, loader_class: type[BaseLoader]) -> None
register_embedder(key: str, embedder_class: type[BaseEmbedder]) -> None
register_generator(key: str, generator_class: type[BaseGenerator]) -> None
register_retriever(key: str, retriever_class: type[BaseRetriever]) -> None
register_vector_store(key: str, store_class: type[BaseVectorStore]) -> None
register_graph_store(key: str, store_class: type[BaseGraphStore]) -> None

from cognity_ai import RAGLibrary

# --- Minimal setup: zero-config, auto-detects available backends ---
rag = RAGLibrary()

# --- Explicit configuration ---
rag = RAGLibrary(
    rag_method="hybrid_graph",
    embedder="openai",
    vector_store="qdrant",
    graph_store="neo4j",
    llm="anthropic",
    openai_api_key="sk-...",
    anthropic_api_key="sk-ant-...",
    neo4j_uri="bolt://localhost:7687",
    neo4j_password="password",
)

# --- Ingest multiple formats ---
rag.ingest("report.pdf")
rag.ingest("data.xlsx")
rag.ingest("slides.pptx")
rag.ingest("photo.jpg")           # OCR via multimodal LLM
count = rag.ingest_dir("./documents/")   # Batch ingest, returns file count

# --- Build knowledge graph communities (run after bulk ingestion) ---
communities = rag.build_communities()
print(f"Built {communities} communities")

# --- Query ---
answer = rag.query("What are the key findings?")
result = rag.query_with_sources("Who is the CEO?", method="hybrid_graph")
print(result["answer"])
for src in result["sources"]:
    print(src["doc_id"], src["source_path"])

# --- Per-query method override ---
summary = rag.query("Summarize all themes", method="microsoft_graphrag")

# --- Retrieval only (no LLM call) ---
results = rag.retrieve("quarterly revenue", top_k=5)
for r in results:
    print(f"[{r.score:.3f}] {r.chunk.text[:120]}...")

# --- Lifecycle management ---
rag.confirm("report_2024")       # Mark as authoritative
rag.deprecate("old_report_2022") # Exclude from retrieval

conflicts = rag.detect_conflicts("Apple Inc")
pruned = rag.prune(threshold=0.4)
print(f"Pruned {pruned} low-confidence items")

report = rag.health_report()
# {"entities": 1243, "relations": 892, "chunks": 3421, "communities": 47,
#  "documents": 28, "deprecated": 2, "confirmed": 5}

# --- Custom plugin: register a loader for a proprietary format ---
from my_loaders import MyCustomLoader
rag.register_loader(".myformat", MyCustomLoader)
rag.ingest("data.myformat")  # Now handled by MyCustomLoader

# --- Custom retriever registered and used immediately ---
from my_retrievers import HybridBM25Retriever
rag.register_retriever("bm25_hybrid", HybridBM25Retriever)
answer = rag.query("latest financials", method="bm25_hybrid")