vector

type Config ¶

type Config struct {
	StoreKind StoreKind
	DB        *DBHandle

	// Deprecated: use StoreKind.
	Backend StoreKind

	Dim                          int
	PgVectorCreateExtension      *bool
	PgVectorIndexMethod          string
	PgVectorHNSWM                int
	PgVectorHNSWEfConstruction   int
	PgVectorIVFFlatLists         int
	QdrantHost                   string
	QdrantPort                   int
	QdrantAPIKey                 string
	QdrantPrefix                 string
	OpenSearchAddresses          []string
	OpenSearchUsername           string
	OpenSearchPassword           string
	OpenSearchAPIKey             string
	OpenSearchIndexPrefix        string
	OpenSearchCreateIndex        *bool
	OpenSearchKNN                bool
	OpenSearchInsecureSkipVerify bool
	OpenSearchBulkSize           int
	ChromaURL                    string
	ChromaAPIKey                 string
	ChromaTenant                 string
	ChromaDatabase               string
	// HNSWDir is the directory for HNSW index files (StoreHNSW only).
	// Defaults to <runtime_root>/data/hnsw via pkg/config helpers.
	HNSWDir string
}

Config selects and tunes a vector store: its kind, the database for the SQL-backed kinds, the dimension of the embeddings, and the settings specific to pgvector, Qdrant, OpenSearch, Chroma and HNSW.

type DBHandle ¶

type DBHandle struct {
	DriverName    string
	DSN           string
	BusyTimeoutMS int
}

DBHandle is the public vector-facing database descriptor. It intentionally avoids leaking the engine's internal DB type while keeping enough information to reopen a compatible low-level handle when needed.

func NewDBHandle(driverName, dsn string) *DBHandle

type Query ¶

type Query = internalvector.Query

Query describes a vector similarity search. Filter applies optional metadata predicates server-side (where supported) or client-side (SQLite / in-memory backends).

Simple equality: {"source_file": "doc.txt"} IN operator: {"source_file": {"$in": ["a.txt", "b.txt"]}}

type Record ¶

type Record = internalvector.Record

Record is one vectorized chunk stored in a vector store.

type SearchResult ¶

type SearchResult = internalvector.SearchResult

SearchResult is one ranked vector search hit.

type Store ¶

type Store = internalvector.Store

Store abstracts the vector storage/search engine used by RAG. Concrete implementations target SQLite, pgvector, Qdrant, Chroma, or any future provider — the rag.Service and knowledge.Service call only this interface.

func NewStore(ctx context.Context, cfg Config) (Store, error)

type StoreKind ¶

type StoreKind = internalvector.StoreKind

StoreKind identifies which vector store implementation to use.