Introducing Bun: A Golang ORM

SQL-first

Bun is a SQL-first Golang ORM for PostgreSQL, MySQL/MariaDB, MSSQL, and SQLite. SQL-first means you can write SQL queries in Go, for example, this Bun query:

var num int
err := db.NewSelect().
    TableExpr("generate_series(1, 3)").
    Where("generate_series = ?", 3).
    Limit(10).
    Scan(ctx, &num)

Generates the following SQL:

SELECT *
FROM generate_series(1, 3)
WHERE generate_series = 123
LIMIT 10

SQL is still there, but Bun helps you generate long queries while protecting against SQL injections thanks to ? placeholders:

Where("id = ?", 123)     // WHERE id = 123
Where("id >= ?", 123)    // WHERE id >= 123
Where("id = ?", "hello") // WHERE id = 'hello'

Where("id IN (?)", bun.In([]int{1, 2, 3})) // WHERE id IN (1, 2, 3)

Where("? = ?", bun.Ident("column"), "value") // WHERE "column" = 'value'

Using Bun, you can write really complex queries, for example, the following Bun query:

regionalSales := db.NewSelect().
	ColumnExpr("region").
	ColumnExpr("SUM(amount) AS total_sales").
	TableExpr("orders").
	GroupExpr("region")

topRegions := db.NewSelect().
	ColumnExpr("region").
	TableExpr("regional_sales").
	Where("total_sales > (SELECT SUM(total_sales) / 10 FROM regional_sales)")

var []items map[string]interface{}

err := db.NewSelect().
	With("regional_sales", regionalSales).
	With("top_regions", topRegions).
	ColumnExpr("region").
	ColumnExpr("product").
	ColumnExpr("SUM(quantity) AS product_units").
	ColumnExpr("SUM(amount) AS product_sales").
	TableExpr("orders").
	Where("region IN (SELECT region FROM top_regions)").
	GroupExpr("region").
	GroupExpr("product").
	Scan(ctx, &items)

Generates the following SQL:

WITH regional_sales AS (
    SELECT region, SUM(amount) AS total_sales
    FROM orders
    GROUP BY region
), top_regions AS (
    SELECT region
    FROM regional_sales
    WHERE total_sales > (SELECT SUM(total_sales)/10 FROM regional_sales)
)
SELECT region,
       product,
       SUM(quantity) AS product_units,
       SUM(amount) AS product_sales
FROM orders
WHERE region IN (SELECT region FROM top_regions)
GROUP BY region, product

Structs and tables

Bun allows you to map Go structs to database tables using struct-based models, for example, the following code:

type Model struct {
    ID        int64 `bun:",pk,autoincrement"`
    Name      string `bun:",notnull"`
    CreatedAt time.Time `bun:",nullzero,default:now()"`
}

err := db.ResetModel(ctx, &Model{})

Generates the following table:

CREATE TABLE "models" (
  "id" BIGSERIAL NOT NULL,
  "name" VARCHAR NOT NULL,
  "created_at" TIMESTAMPTZ DEFAULT now(),
  PRIMARY KEY ("id"),
)

You can then select/insert/update/delete rows using Go structs:

model := new(Model)
err := db.NewSelect().Model().Where("id = ?", 123).Scan(ctx)

model.ID = 0
res, err := db.NewInsert().Model(model).Exec(ctx)

res, err := db.NewUpdate().
    Model(model).
    Set("name = ?", "updated name").
    WherePK().
    Exec(ctx)

res, err := db.NewDelete().Model(model).WherePK().Exec(ctx)

See Bun documentation for details.

Golang ORM

So what about the Golang ORM part? Bun allows you to define common table relations using Go structs, for example, here is how you can define Author belongs to Book relation:

type Book struct {
	ID		 int64
	AuthorID int64
	Author	 Author `bun:"rel:belongs-to,join:author_id=id"`
}

type Author struct {
	ID int64
}

And then use Relation method to join tables:

err := db.NewSelect().
	Model(book).
	Relation("Author").
	Where("id = ?", 123).
	Scan(ctx)

SELECT
  "book"."id", "book"."title", "book"."text",
  "author"."id" AS "author__id", "author"."name" AS "author__name"
FROM "books"
LEFT JOIN "users" AS "author" ON "author"."id" = "book"."author_id"
WHERE id = 1

See ORM: Table relationships for details.

Connecting to a database

Bun works on top of database/SQL and supports PostgreSQL, MySQL/MariaDB, MSSQL, and SQLite.

To connect to a PostgreSQL database:

import (
	"github.com/uptrace/bun"
	"github.com/uptrace/bun/dialect/pgdialect"
    "github.com/uptrace/bun/driver/pgdriver"
)

dsn := "postgres://postgres:@localhost:5432/test?sslmode=disable"
sqldb := sql.OpenDB(pgdriver.NewConnector(pgdriver.WithDSN(dsn)))

db := bun.NewDB(sqldb, pgdialect.New())

To connect to a MySQL database:

import (
	"github.com/uptrace/bun"
	"github.com/uptrace/bun/dialect/mysqldialect"
	_ "github.com/go-sql-driver/mysql"
)

sqldb, err := sql.Open("mysql", "root:pass@/test")
if err != nil {
	panic(err)
}

db := bun.NewDB(sqldb, mysqldialect.New())

To log all executed queries, you can install bundebug plugin:

import "github.com/uptrace/bun/extra/bundebug"

db.AddQueryHook(bundebug.NewQueryHook(
	bundebug.WithVerbose(true), // log everything
))

Executing queries

Once you have a model, you can start executing queries:

// Select a user by a primary key.
user := new(User)
err := db.NewSelect().Model(user).Where("id = ?", 1).Scan(ctx)

// Select first 10 users.
var users []User
err := db.NewSelect().Model(&users).OrderExpr("id ASC").Limit(10).Scan(ctx)

When it comes to scanning query results, Bun is very flexible and allows scanning into structs:

user := new(User)
err := db.NewSelect().Model(user).Limit(1).Scan(ctx)

Into scalars:

var id int64
var name string
err := db.NewSelect().Model((*User)(nil)).Column("id", "name").Limit(1).Scan(ctx, &id, &name)

Into a map[string]interface{}:

var m map[string]interface{}
err := db.NewSelect().Model((*User)(nil)).Limit(1).Scan(ctx, &m)

And into slices of the types above:

var users []User
err := db.NewSelect().Model(&users).Limit(1).Scan(ctx)

var ids []int64
var names []string
err := db.NewSelect().Model((*User)(nil)).Column("id", "name").Limit(1).Scan(ctx, &ids, &names)

var ms []map[string]interface{}
err := db.NewSelect().Model((*User)(nil)).Scan(ctx, &ms)

You can also return results from insert/update/delete queries and scan them too:

var ids []int64
res, err := db.NewDelete().Model((*User)(nil)).Returning("id").Exec(ctx, &ids)

What's next?

To get started, see the documentation and run examples.

Bun comes with many plugins including OpenTelemetry instrumentation that enables distributed tracing and metrics.

Using tracing, you can monitor performance using one of the open-source tracing tools that work with OpenTelemetry. Many DataDog competitors also support OpenTelemetry.

Besides, you can export metrics to Prometheus and visualize them using Grafana or a popular alternative.