embedding

type ContentResolver interface {
	ResolveContent(ctx context.Context, collection, id string) (string, error)
}

type EmbedRequest struct {
	ID         string
	Collection string
	Content    string
	Metadata   map[string]string
}

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

type EmbeddingProvider interface {
	// ID returns the unique identifier for this provider.
	ID() string
	// Embed generates embeddings for the given texts in a single batch.
	Embed(ctx context.Context, texts []string) ([][]float32, error)
	// Dimensions returns the dimensionality of the generated embeddings.
	Dimensions() int
}

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

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

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

type ProviderConfig struct {
	// Provider selects the embedding backend: "openai", "google", or "local".
	Provider string
	// Model is the embedding model identifier.
	Model string
	// Dimensions is the embedding vector dimensionality.
	Dimensions int
	// APIKey is required for openai and google providers.
	APIKey string
	// BaseURL is used by the local provider (Ollama endpoint).
	BaseURL string
}

type RAGResult struct {
	Collection string
	SourceID   string
	Content    string
	Distance   float32
}

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

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

type RetrieveOptions struct {
	// Collections to search (empty means all).
	Collections []string
	// Maximum results to return.
	Limit int
	// SessionKey filters for session-specific results.
	SessionKey string
	// MaxDistance is the maximum cosine distance for results (0.0 = disabled).
	MaxDistance float32
}

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

type SearchOptions struct {
	// MetadataFilter post-filters results by metadata key-value pairs.
	MetadataFilter map[string]string
}

type SearchResult struct {
	ID         string
	Collection string
	Distance   float32
	Metadata   map[string]string
}

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

type VectorRecord struct {
	// ID is the source entity identifier.
	ID string
	// Collection groups records (e.g., "knowledge", "observation", "reflection", "learning").
	Collection string
	// Embedding is the vector representation.
	Embedding []float32
	// Metadata holds filterable key-value pairs (e.g., session_key, category).
	Metadata map[string]string
}

type VectorStore interface {
	// Upsert inserts or replaces vector records.
	Upsert(ctx context.Context, records []VectorRecord) error
	// Search finds the most similar vectors in a collection.
	// Pass nil for opts to use default behavior (no filtering).
	Search(ctx context.Context, collection string, query []float32, limit int, opts *SearchOptions) ([]SearchResult, error)
	// Delete removes vectors by collection and source IDs.
	Delete(ctx context.Context, collection string, ids []string) error
	// Close releases any resources held by the store.
	Close() error
}