pgvector-zig

pgvector examples for Zig

Supports pg.zig and libpq

Getting Started

Follow the instructions for your database library:

• pg.zig
• libpq

Or check out some examples:

• Embeddings with OpenAI
• Binary embeddings with Cohere
• Hybrid search with Ollama (Reciprocal Rank Fusion)
• Sparse search with Text Embeddings Inference

pg.zig

Enable the extension

_ = try conn.exec("CREATE EXTENSION IF NOT EXISTS vector", .{});

Create a table

_ = try conn.exec("CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3))", .{});

Insert vectors

const embedding1 = [_]f32{ 1, 2, 3 };
const embedding2 = [_]f32{ 4, 5, 6 };
_ = try conn.exec("INSERT INTO items (embedding) VALUES ($1::float4[]), ($2::float4[])", .{ embedding1, embedding2 });

Get the nearest neighbors

const embedding3 = [_]f32{ 3, 1, 2 };
var result = try conn.query("SELECT id FROM items ORDER BY embedding <-> $1::float4[]::vector LIMIT 5", .{embedding3});

Add an approximate index

_ = try conn.exec("CREATE INDEX ON items USING hnsw (embedding vector_l2_ops)", .{});
// or
_ = try conn.exec("CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)", .{});

Use vector_ip_ops for inner product and vector_cosine_ops for cosine distance

See a full example

libpq

Import libpq

const pg = @cImport({
    @cInclude("libpq-fe.h");
});

Enable the extension

const res = pg.PQexec(conn, "CREATE EXTENSION IF NOT EXISTS vector");

Create a table

const res = pg.PQexec(conn, "CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3))");

Insert vectors

const paramValues = [2:0][*c]const u8{ "[1,2,3]", "[4,5,6]" };
const res = pg.PQexecParams(conn, "INSERT INTO items (embedding) VALUES ($1), ($2)", 2, null, &paramValues, null, null, 0);

Get the nearest neighbors

const paramValues = [1:0][*c]const u8{"[3,1,2]"};
const res = pg.PQexecParams(conn, "SELECT * FROM items ORDER BY embedding <-> $1 LIMIT 5", 1, null, &paramValues, null, null, 0);

Add an approximate index

const res = pg.PQexec(conn, "CREATE INDEX ON items USING hnsw (embedding vector_l2_ops)");
// or
const res = pg.PQexec(conn, "CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)");

Use vector_ip_ops for inner product and vector_cosine_ops for cosine distance

See a full example

Contributing

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

• Report bugs
• Fix bugs and submit pull requests
• Write, clarify, or fix documentation
• Suggest or add new features

To get started with development:

git clone https://github.com/pgvector/pgvector-zig.git
cd pgvector-zig
createdb pgvector_zig_test
zig build
zig-out/bin/pg
zig-out/bin/libpq

Specify the path to libpq if needed:

zig build --search-prefix /opt/homebrew/opt/libpq

To run an example:

createdb pgvector_example
examples/openai