datastore

Documentation

Index

• func WithAdditionalFields(fields []string) func(*Option)
• func WithLimit(limit int64) func(*Option)
• func WithSimilarity(similarity float32) func(*Option)
• type DataStore
• type Option
• type Retriever
• type VectorStore

Functions

func WithAdditionalFields

func WithAdditionalFields(fields []string) func(*Option)

WithAdditionalFields set the additional fields to query Some datastore has different way to handle non existent field, some return empty some return error

func WithLimit

func WithLimit(limit int64) func(*Option)

WithLimit set the limit of the result

func WithSimilarity

func WithSimilarity(similarity float32) func(*Option)

WithSimilarity set the minimum similarity score Different datastore has different way to handle similarity score

Types

type DataStore

type DataStore interface {
	// Search using a query string
	Search(ctx context.Context, indexName string, query string, options ...func(*Option)) ([]document.Document, error)
	// AddText store text to vector index
	// it will also store embedding of the text using specified embedding model
	// If the index doesnt exist, it will create one with default schema
	AddText(ctx context.Context, indexName string, input string) (err error)
	// AddDocuments store a batch of documents to vector index
	// it will also store embedding of the document using specified embedding model
	// If the index doesnt exist, it will create one with default schema
	AddDocuments(ctx context.Context, indexName string, documents []document.Document) (batchErr []error, err error)
	// DeleteIndex drop the index
	DeleteIndex(ctx context.Context, indexName string) (err error)
}

DataStore is the interface for storing and retrieving data

type Option

type Option struct {
	Limit            int64    // max result to return
	Similarity       float32  // minimum similarity score
	AdditionalFields []string // list of fields to return, Text field is always returned part of the Document
}

Option give way to pass additional parameter to the datastore

type Retriever

type Retriever interface {
	// Search using a query string
	Search(ctx context.Context, indexName string, query string, options ...func(*Option)) ([]document.Document, error)
}

Retriever is the interface for retrieving data can be used to interact with read only data source like get API DataStore and VectorStore are interface compatible with retriever

type VectorStore

type VectorStore interface {
	DataStore
	//SearchVector by providing the vector from embedding function
	SearchVector(ctx context.Context, indexName string, vector []float32, options ...func(*Option)) ([]document.Document, error)
}

VectorStore is the interface for storing and retrieving vector data