Awesome Open Source
Awesome Open Source

koochooloo 🌰

a Persian word which means "small" or "li'l". It is often used to refer to a girl when flirting, (with the meaning, li'l girl)

Urban Dictionary

GitHub Workflow Status Codecov Docker Image Size (tag) Docker Pulls PkgGoDev


Here is a mini project for shortening your URLs. This sweet project shows how to write a simple lovely Golang's project that contains Database, Configuration, and, etc. You can use this project as a guidance to write your ReST applications. This project tries to be strongly typed, easy to read and easy to maintain therefore there is no global variable, init function and etc. We have used the singular name for package as a de-facto standard, and the structure is based on project-layout.

I want to dedicate this project to my love ❤️.

The goal is have a project that you can add features into it easily and without struggling with the code base. Each package works independently from other packages and you can find easily what you need.



First of all, cmd package contains the binaries of this project with use of cobra. It is good to have a simple binary for database migrations that can be run on initiation phase of project. Each binary has its main.go in its package and registers itself with a Register function. In the root.go of cmd configuration and other shared things are initiated. Here is an example for register function:

// Register server command.
func Register(root *cobra.Command, cfg config.Config) {
			Use:   "server",
			Short: "Run server to serve the requests",
			Run: func(cmd *cobra.Command, args []string) {

Again each command registers its flag by itself so we have sepration from other commands. Sometimes we need to have shared flags between commands, then it is better to have them in config. For the later case, koanf can help us with the structure as below:

func Register(fs *pflag.FlagSet) {
		"url", "u",
		fmt.Sprintf("nats server url(s) e.g. %s", nats.DefaultURL),

This function register shared flags and then we load configuration based on them with the following function:

k := koanf.New(".")
if err := k.Load(posflag.Provider(fs, ".", k), nil); err != nil {
	log.Errorf("error loading config.yml: %s", err)

if err := k.Unmarshal("", &instance); err != nil {
	log.Fatalf("error unmarshalling config: %s", err)


The main part of each application is its configuration. There are many ways for having configuration in the project from configuration file to environment variables. koanf has all of them. The main points here are:

  • having a defined and typed structure for configuration
  • don't use global configuration. each module has its configuration defined in config module and it will pass to it in its initiation.
  • print loaded configuration at startup so everyone can validate the applied configuration.

P.S. koanf is way better than viper for having typed configuration. By typed configuration I mean you have a defined structure for configuration and then load configuration from many sources into it.


There is a db package that is responsible for connecting to the database. This package uses the database configuration that is defined in config module and create a database instance. It is good to ping your database here to have fully confident to your database instance. Also for having an insight at database health you can call this ping function periodically and report its result with metrics.


Project models are defined in model package. These models are used internally but the can be used in response or request package. There is no structure for communicating with database in this package.


Stores are responsible for commnunicating with database to store or retrieve models. Stores are interface and there is an SQL and moked version for them. SQL model is used in main code and mocked is used for tests. Please note that the tests for SQL stores are touchy and are done with actual database.


HTTP handler are defined in handler package. Echo is an awesome HTTP framework that has eveything you need. Each handler has its structure with a Register method that registers its route into a given route group. Route group is a concept from Echo framework for grouping routes under a specific parent path. Each handler has what it needs into its structure. Handler structure are created in main.go then register on their group.

type Healthz struct {

// Handle shows server is up and running.
func (h Healthz) Handle(c echo.Context) error {
	return c.NoContent(http.StatusNoContent)

// Register registers the routes of healthz handler on given echo group.
func (h Healthz) Register(g *echo.Group) {
	g.GET("/healthz", h.Handle)


All metrics are gathered using prometheus. Each package has its metric.go that defines a structure contains the metrics and have methods for changing them. For migrating from prmotheus you just need to change metric.go. Metrics aren't global and they created for each instance seperately and with the following code there is no issue with duplicate registration for prometheus metrics.

// my_mertic is a prometheus.Counter

if err := prometheus.Register(my_metric); err != nil {
    var are prometheus.AlreadyRegisteredError
    if ok := errors.As(err, &are); ok {
        my_metric = are.ExistingCollector.(prometheus.Counter)
    } else {

For having better controller on metrics endpoint there is another HTTP server that is defined in metric package for monitoring.


It is good to have sperated pakcages for requests and responses. These packages also contain validation logic. One of the good validation pakcages in Go is ozzo-validator. After providing validate method, after getting request you can validate it with its method with ease.


Logging one the most important part of application. At the beginning there is no need to have something more than simple stdout logs. But in the future you need to strcuture you logs and ship them into an aggregation system because when your system grows detecting issues from text logs will be inpossible.

zap is one the best logger for structure logging. zap forces you to pass it into your child module and you also name loggers with Named method. By using the named logger you can easily find you module logs in your log aggregator.

Up and Running

This project only requires MongoDB, and you can run it with provided docker-compose.

go build
docker-compose up -d
curl -X POST -d '{"url": ""}' -H 'Content-Type: application/json'
curl -L

Load Testing

    checks.....................: 99.83%  2995   5
    data_received..............: 2.0 MB 64 kB/s
    data_sent..................: 521 kB 17 kB/s
    group_duration.............: avg=649.18ms min=153.18s med=265.45ms max=30.95s   p(90)=1.61s    p(95)=2.06s
    http_req_blocked...........: avg=14.12ms  min=0s       med=3s      max=1.65s    p(90)=13s     p(95)=147.04s
    http_req_connecting........: avg=6.23ms   min=0s       med=0s       max=1.36s    p(90)=0s       p(95)=0s
    http_req_duration..........: avg=272.98ms min=0s       med=127.99ms max=4.81s    p(90)=830.93ms p(95)=1.29s
    http_req_receiving.........: avg=125.23s min=0s       med=60s     max=11.21ms  p(90)=228s    p(95)=363s
    http_req_sending...........: avg=50.78s  min=0s       med=22s     max=7.28ms   p(90)=86s     p(95)=138s
    http_req_tls_handshaking...: avg=7.86ms   min=0s       med=0s       max=653.63ms p(90)=0s       p(95)=0s
    http_req_waiting...........: avg=272.8ms  min=0s       med=127.71ms max=4.81s    p(90)=830.87ms p(95)=1.29s
    http_reqs..................: 4000   129.093962/s
    iteration_duration.........: avg=1.29s    min=142.34ms med=1.04s    max=30.97s   p(90)=2.18s    p(95)=2.64s
    iterations.................: 1000   32.273491/s
    vus........................: 100    min=100 max=100
    vus_max....................: 100    min=100 max=100

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
Go (193,976
Golang (32,131
Learning By Doing (1,050
Url Shortener (838
Learn To Code (529
Golang Application (466
Golang Examples (192
Related Projects