I'm new at Go programming, and I'm wondering: what is the preferred way to handle configuration parameters for a Go program (the kind of stuff one might use properties files or ini files for, in other contexts)?
This question is related to
go
configuration-files
have a look at gonfig
// load
config, _ := gonfig.FromJson(myJsonFile)
// read with defaults
host, _ := config.GetString("service/host", "localhost")
port, _ := config.GetInt("service/port", 80)
test, _ := config.GetBool("service/testing", false)
rate, _ := config.GetFloat("service/rate", 0.0)
// parse section into target structure
config.GetAs("service/template", &template)
I tried JSON. It worked. But I hate having to create the struct of the exact fields and types I might be setting. To me that was a pain. I noticed it was the method used by all the configuration options I could find. Maybe my background in dynamic languages makes me blind to the benefits of such verboseness. I made a new simple config file format, and a more dynamic-ish lib for reading it out.
https://github.com/chrisftw/ezconf
I am pretty new to the Go world, so it might not be the Go way. But it works, it is pretty quick, and super simple to use.
I wrote a simple ini config library in golang.
goroutine-safe, easy to use
package cfg
import (
"testing"
)
func TestCfg(t *testing.T) {
c := NewCfg("test.ini")
if err := c.Load() ; err != nil {
t.Error(err)
}
c.WriteInt("hello", 42)
c.WriteString("hello1", "World")
v, err := c.ReadInt("hello", 0)
if err != nil || v != 42 {
t.Error(err)
}
v1, err := c.ReadString("hello1", "")
if err != nil || v1 != "World" {
t.Error(err)
}
if err := c.Save(); err != nil {
t.Error(err)
}
}
===================Update=======================
Recently I need an INI parser with section support, and I write a simple package:
github.com/c4pt0r/cfg
u can parse INI like using "flag" package:
package main
import (
"log"
"github.com/c4pt0r/ini"
)
var conf = ini.NewConf("test.ini")
var (
v1 = conf.String("section1", "field1", "v1")
v2 = conf.Int("section1", "field2", 0)
)
func main() {
conf.Parse()
log.Println(*v1, *v2)
}
Another option is to use TOML, which is an INI-like format created by Tom Preston-Werner. I built a Go parser for it that is extensively tested. You can use it like other options proposed here. For example, if you have this TOML data in something.toml
Age = 198
Cats = [ "Cauchy", "Plato" ]
Pi = 3.14
Perfection = [ 6, 28, 496, 8128 ]
DOB = 1987-07-05T05:45:00Z
Then you can load it into your Go program with something like
type Config struct {
Age int
Cats []string
Pi float64
Perfection []int
DOB time.Time
}
var conf Config
if _, err := toml.DecodeFile("something.toml", &conf); err != nil {
// handle error
}
I usually use JSON for more complicated data structures. The downside is that you easily end up with a bunch of code to tell the user where the error was, various edge cases and what not.
For base configuration (api keys, port numbers, ...) I've had very good luck with the gcfg package. It is based on the git config format.
From the documentation:
Sample config:
; Comment line
[section]
name = value # Another comment
flag # implicit value for bool is true
Go struct:
type Config struct {
Section struct {
Name string
Flag bool
}
}
And the code needed to read it:
var cfg Config
err := gcfg.ReadFileInto(&cfg, "myconfig.gcfg")
It also supports slice values, so you can allow specifying a key multiple times and other nice features like that.
I have started using Gcfg which uses Ini-like files. It's simple - if you want something simple, this is a good choice.
Here's the loading code I am using currently, which has default settings and allows command line flags (not shown) that override some of my config:
package util
import (
"code.google.com/p/gcfg"
)
type Config struct {
Port int
Verbose bool
AccessLog string
ErrorLog string
DbDriver string
DbConnection string
DbTblPrefix string
}
type configFile struct {
Server Config
}
const defaultConfig = `
[server]
port = 8000
verbose = false
accessLog = -
errorLog = -
dbDriver = mysql
dbConnection = testuser:TestPasswd9@/test
dbTblPrefix =
`
func LoadConfiguration(cfgFile string, port int, verbose bool) Config {
var err error
var cfg configFile
if cfgFile != "" {
err = gcfg.ReadFileInto(&cfg, cfgFile)
} else {
err = gcfg.ReadStringInto(&cfg, defaultConfig)
}
PanicOnError(err)
if port != 0 {
cfg.Server.Port = port
}
if verbose {
cfg.Server.Verbose = true
}
return cfg.Server
}
Just use standard go flags with iniflags.
Standard go flags have the following benefits:
The only drawback standard go flags have - is management problems when the number of flags used in your app becomes too large.
Iniflags elegantly solves this problem: just modify two lines in your main package and it magically gains support for reading flag values from ini file. Flags from ini files can be overriden by passing new values in command-line.
See also https://groups.google.com/forum/#!topic/golang-nuts/TByzyPgoAQE for details.
You might also be interested in go-libucl, a set of Go bindings for UCL, the Universal Configuration Language. UCL is a bit like JSON, but with better support for humans: it supports comments and human-readable constructs like SI multipliers (10k, 40M, etc.) and has a little bit less boilerplate (e.g., quotes around keys). It's actually pretty close to the nginx configuration file format, if you're already familiar with that.
I agree with nemo and I wrote a little tool to make it all real easy.
bitbucket.org/gotamer/cfg is a json configuration package
See doc.go for an example
Use toml like this article Reading config files the Go way
Viper is a golang configuration management system that works with JSON, YAML, and TOML. It looks pretty interesting.
https://github.com/spf13/viper and https://github.com/zpatrick/go-config are a pretty good libraries for configuration files.
Source: Stackoverflow.com