I have a Go project using goose for Mysql migrations. I would like to bind the migrations to the package executable so that the executable can be deployed and used independently
Install
go get -u github.com/pressly/goose/cmd/goose
Make app. I base it on examplemain.go
and add run
option. Suppose your project is located at github.com/user/project
:
package main
import (
"database/sql"
"flag"
"log"
"os"
"github.com/pressly/goose"
// Init DB drivers. -- here I recommend remove unnecessary - but it's up to you
_ "github.com/go-sql-driver/mysql"
_ "github.com/lib/pq"
_ "github.com/mattn/go-sqlite3"
_ "github.com/ziutek/mymysql/godrv"
// here our migrations will live -- use your path
_ "github.com/user/project/migrations"
)
var (
flags = flag.NewFlagSet("goose", flag.ExitOnError)
dir = flags.String("dir", ".", "directory with migration files")
)
func main() {
flags.Usage = usage
flags.Parse(os.Args[1:])
args := flags.Args()
//////
if len(args) > 1 && args[0] == "run" {
log.Printf("PROGRAM RUN\n") //
.....
os.Exit(0)
}
if len(args) > 1 && args[0] == "create" {
if err := goose.Run("create", nil, *dir, args[1:]...); err != nil {
log.Fatalf("goose run: %v", err)
}
return
}
if len(args) < 3 {
flags.Usage()
return
}
if args[0] == "-h" || args[0] == "--help" {
flags.Usage()
return
}
driver, dbstring, command := args[0], args[1], args[2]
switch driver {
case "postgres", "mysql", "sqlite3", "redshift":
if err := goose.SetDialect(driver); err != nil {
log.Fatal(err)
}
default:
log.Fatalf("%q driver not supported\n", driver)
}
switch dbstring {
case "":
log.Fatalf("-dbstring=%q not supported\n", dbstring)
default:
}
if driver == "redshift" {
driver = "postgres"
}
db, err := sql.Open(driver, dbstring)
if err != nil {
log.Fatalf("-dbstring=%q: %v\n", dbstring, err)
}
arguments := []string{}
if len(args) > 3 {
arguments = append(arguments, args[3:]...)
}
if err := goose.Run(command, db, *dir, arguments...); err != nil {
log.Fatalf("goose run: %v", err)
}
}
func usage() {
log.Print(usagePrefix)
flags.PrintDefaults()
log.Print(usageCommands)
}
var (
usagePrefix = `Usage: goose [OPTIONS] DRIVER DBSTRING COMMAND
Drivers:
postgres
mysql
sqlite3
redshift
Examples:
goose sqlite3 ./foo.db status
goose sqlite3 ./foo.db create init sql
goose sqlite3 ./foo.db create add_some_column sql
goose sqlite3 ./foo.db create fetch_user_data go
goose sqlite3 ./foo.db up
goose postgres "user=postgres dbname=postgres sslmode=disable" status
goose mysql "user:password@/dbname?parseTime=true" status
goose redshift "postgres://user:password@qwerty.us-east-1.redshift.amazonaws.com:5439/db"
status
Options:
`
usageCommands = `
Commands:
up Migrate the DB to the most recent version available
up-to VERSION Migrate the DB to a specific VERSION
down Roll back the version by 1
down-to VERSION Roll back to a specific VERSION
redo Re-run the latest migration
status Dump the migration status for the current DB
version Print the current version of the database
create NAME [sql|go] Creates new migration file with next version
`
)
Create folder for migrations:
mkdir migrations && cd migrations
Create first migrations. We will use go
-style migrations:
goose mysql "user:password@/dbname?parseTime=true" create init go
You'll get a file 00001_init.go
with Go code. Migrations are baked in it as SQL-commands. Just edit them as you need.
Then go to the main folder and build the application:
cd ..
go build -v -o myapp *.go
You'll get a file myapp
with all the migrations baked in it. To check move it to some other place, for example to /tmp
folder, and run from there:
./myapp mysql "user:password@/dbname?parseTime=true" status
Run your app:
./myapp run
You have single file which can be used as a migration tool so as a working application itself. All the migration are buil-it. In source code they are stored in a subpackage migrations
- so it's easy to edit.
If you're already using Goose, one option would be to write the migrations in Go instead of SQL. Based on the Go migrations example in the Goose repo, when you build the goose
binary here, it will bundle all the *.go
ones into the binary.
This was the output after I built the example and removed all files except the binary itself. The Go-based migration was embedded:
2017/10/31 11:22:31 Applied At Migration
2017/10/31 11:22:31 =======================================
2017/10/31 11:22:31 Mon Jun 19 21:56:00 2017 -- 00002_rename_root.go
If you're looking to use SQL-based migrations but don't want to take on additional dependencies, you could embed your SQL-based migrations into *.go
files as string constants, then, starting from the go-migrations
example, add an init phase to main.go
that writes them out to the current directory before proceeding.
If you use Docker you may put folder to the image.
If not projects like https://github.com/rakyll/statik or https://github.com/jteeuwen/go-bindata can help.