Struct PgVector

pub struct PgVector { /* private fields */ }

Available on crate feature pgvector only.

Represents a Pgvector client with configuration options.

This struct is used to interact with the Pgvector vector database, providing methods to manage vector collections, store data, and ensure efficient searches. The client can be cloned with low cost as it shares connections.

Implementations

impl PgVector

pub fn generate_create_table_sql(&self) -> Result<String>

Generates a SQL statement to create a table for storing vector embeddings.

The table will include columns for an ID, chunk data, metadata, and a vector embedding.

Returns

• The generated SQL statement.

Errors

• Returns an error if the table name is invalid or if vector_size is not configured.

pub fn create_index_sql(&self) -> Result<String>

Generates the SQL statement to create an HNSW index on the vector column.

Errors

Returns an error if:

• No vector field is found in the table configuration.
• The table name or field name is invalid.

pub async fn store_nodes(&self, nodes: &[Node]) -> Result<()>

Stores a list of nodes in the database using an upsert operation.

Arguments

• nodes - A slice of Node objects to be stored.

Returns

• Result<()> - Ok if all nodes are successfully stored, Err otherwise.

Errors

This function will return an error if:

• The database connection pool is not established.
• Any of the SQL queries fail to execute due to schema mismatch, constraint violations, or connectivity issues.
• Committing the transaction fails.

pub fn get_vector_column_name(&self) -> Result<String>

Retrieves the name of the vector column configured in the schema.

Returns

• Ok(String) - The name of the vector column if exactly one is configured.

Errors

• Error::NoEmbedding - If no vector field is configured in the schema.
• Error::MultipleEmbeddings - If multiple vector fields are configured in the schema.

impl PgVector

pub fn normalize_field_name(field: &str) -> String

impl PgVector

pub async fn pool_get_or_initialize(&self) -> Result<&PgPool>

Returns a reference to the PgPool if it is already initialized, or creates and initializes it if it is not.

Errors

This function will return an error if pool creation fails.

impl PgVector

pub fn builder() -> PgVectorBuilder

Creates a new instance of PgVectorBuilder with default settings.

Returns

A new PgVectorBuilder.

pub async fn get_pool(&self) -> Result<&PgPool>

Retrieves a connection pool for PostgreSQL.

This function returns the connection pool used for interacting with the PostgreSQL database. It fetches the pool from the PgDBConnectionPool struct.

Returns

A Result that, on success, contains the PgPool representing the database connection pool. On failure, an error is returned.

Errors

This function will return an error if it fails to retrieve the connection pool, which could occur if the underlying connection to PostgreSQL has not been properly established.

pub fn get_table_name(&self) -> &str

Trait Implementations

impl Clone for PgVector

fn clone(&self) -> PgVector

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for PgVector

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Persist for PgVector

fn setup<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

fn store<'life0, 'async_trait>( &'life0 self, node: Node, ) -> Pin<Box<dyn Future<Output = Result<Node>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

fn batch_store<'life0, 'async_trait>( &'life0 self, nodes: Vec<Node>, ) -> Pin<Box<dyn Future<Output = IndexingStream> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

fn batch_size(&self) -> Option<usize>

fn name(&self) -> &'static str

impl Retrieve<CustomStrategy<QueryBuilder<'static, Postgres>>> for PgVector

fn retrieve<'life0, 'life1, 'async_trait>( &'life0 self, search_strategy: &'life1 CustomStrategy<QueryBuilder<'static, Postgres>>, query: Query<Pending>, ) -> Pin<Box<dyn Future<Output = Result<Query<Retrieved>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

fn name(&self) -> &'static str

impl Retrieve<SimilaritySingleEmbedding<String>> for PgVector

fn retrieve<'life0, 'life1, 'async_trait>( &'life0 self, search_strategy: &'life1 SimilaritySingleEmbedding<String>, query_state: Query<Pending>, ) -> Pin<Box<dyn Future<Output = Result<Query<Retrieved>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

fn name(&self) -> &'static str

impl Retrieve<SimilaritySingleEmbedding> for PgVector

fn retrieve<'life0, 'life1, 'async_trait>( &'life0 self, search_strategy: &'life1 SimilaritySingleEmbedding, query: Query<Pending>, ) -> Pin<Box<dyn Future<Output = Result<Query<Retrieved>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

fn name(&self) -> &'static str