pgvector 0.1.4

pgvector support for Rust
Documentation

pgvector-rust

pgvector support for Rust

Supports Rust-Postgres, SQLx, and Diesel

Build Status

Getting Started

Follow the instructions for your database library:

Rust-Postgres

Add this line to your application’s Cargo.toml under [dependencies]:

pgvector = { version = "0.1", features = ["postgres"] }

Create a vector from a Vec<f32>

let vec = pgvector::Vector::from(vec![1.0, 2.0, 3.0]);

Insert a vector

client.execute("INSERT INTO table (column) VALUES ($1)", &[&vec])?;

Get the nearest neighbor

let row = client.query_one("SELECT * FROM table ORDER BY column <-> $1 LIMIT 1", &[&vec])?;

Retrieve a vector

let row = client.query_one("SELECT column FROM table LIMIT 1", &[])?;
let vec: pgvector::Vector = row.get(0);

Use Option if the value could be NULL

let res: Option<pgvector::Vector> = row.get(0);

SQLx

Add this line to your application’s Cargo.toml under [dependencies]:

pgvector = { version = "0.1", features = ["sqlx"] }

Create a vector from a Vec<f32>

let vec = pgvector::Vector::from(vec![1.0, 2.0, 3.0]);

Insert a vector

sqlx::query("INSERT INTO table (column) VALUES ($1)").bind(vec).execute(&pool).await?;

Get the nearest neighbors

let rows = sqlx::query("SELECT * FROM table ORDER BY column <-> $1 LIMIT 1")
    .bind(vec).fetch_all(&pool).await?;

Retrieve a vector

let row = sqlx::query("SELECT column FROM table LIMIT 1").fetch_one(&pool).await?;
let vec: pgvector::Vector = row.try_get("column")?;

Diesel

Add this line to your application’s Cargo.toml under [dependencies]:

pgvector = { version = "0.1", features = ["diesel"] }

And add this line to your application’s diesel.toml under [print_schema]:

import_types = ["diesel::sql_types::*", "pgvector::sql_types::*"]

Create a migration

diesel migration generate create_vector_extension

with up.sql:

CREATE EXTENSION vector

and down.sql:

DROP EXTENSION vector

Run the migration

diesel migration run

You can now use the vector type in future migrations

CREATE TABLE items (
  factors VECTOR(3)
)

For models, use:

pub struct Item {
    pub factors: Option<pgvector::Vector>
}

Create a vector from a Vec<f32>

let factors = pgvector::Vector::from(vec![1.0, 2.0, 3.0]);

Insert a vector

let new_item = Item {
    factors: Some(factors)
};

diesel::insert_into(items::table)
    .values(&new_item)
    .get_result::<Item>(&conn)?;

Get the nearest neighbors

use pgvector::VectorExpressionMethods;

let neighbors = items::table
    .order(items::factors.l2_distance(factors))
    .limit(5)
    .load::<Item>(&conn)?;

Also supports max_inner_product and cosine_distance

Get the distances

let distances = items::table
    .select(items::factors.l2_distance(factors))
    .load::<f64>(&conn)?;

Add an approximate index in a migration

CREATE INDEX my_index ON items USING ivfflat (factors vector_l2_ops)

Use vector_ip_ops for inner product and vector_cosine_ops for cosine distance

Reference

Convert a vector to a Vec<f32>

let f32_vec: Vec<f32> = vec.into();

History

View the changelog

Contributing

Everyone is encouraged to help improve this project. Here are a few ways you can help:

To get started with development:

git clone https://github.com/ankane/pgvector-rust.git
cd pgvector-rust
createdb pgvector_rust_test
cargo test --all-features