pgvector

Documentation

Overview

Package pgvector provides a MemoryStore backed by PostgreSQL with the pgvector extension. Register type: "pgvector".

Metadata keys:

dsn           — PostgreSQL DSN (required)
table         — table name, default daimon_documents
embedding_url — OpenAI-compatible endpoint for embedding generation
dimensions    — embedding dimensions, default 1536

Index

• type Store
  • func New(cfg memory.StoreConfig) (*Store, error)
  • func (s *Store) Delete(ctx context.Context, id string) error
  • func (s *Store) Query(ctx context.Context, query string, topK int) ([]memory.Result, error)
  • func (s *Store) Upsert(ctx context.Context, id, content string, metadata map[string]string) (string, error)

Types

type Store

type Store struct {
	// contains filtered or unexported fields
}

Store uses PostgreSQL + pgvector for vector similarity search.

func New

func New(cfg memory.StoreConfig) (*Store, error)

New creates a Store and initialises the table and index.

func (*Store) Delete

func (s *Store) Delete(ctx context.Context, id string) error

func (*Store) Query

func (s *Store) Query(ctx context.Context, query string, topK int) ([]memory.Result, error)

func (*Store) Upsert

func (s *Store) Upsert(ctx context.Context, id, content string, metadata map[string]string) (string, error)