yaml

YAML marshaling and unmarshaling support for Go

Fork

This fork is an improved version of the invopop/yaml package, designed to include line and column location information for YAML elements during unmarshalling.

Origin tracking uses a two-pass approach:

1. UnmarshalWithOriginTree decodes the YAML and extracts __origin__ metadata injected by the underlying oasdiff/yaml3 decoder, returning an *OriginTree alongside the decoded struct.
2. The caller walks the OriginTree to apply file/line/column information to the decoded Go structs.

To decode without origin tracking, use Unmarshal as usual. To decode with origin tracking:

tree, err := yaml.UnmarshalWithOriginTree(data, &v, yaml.OriginOpt{Enabled: true, File: "myfile.yaml"})

The returned *OriginTree mirrors the YAML document structure. Each node holds a compact []any sequence with the file, key name, line, column, and locations of scalar fields and sequence items within that mapping. When Enabled is false, nil is returned for the tree with no overhead.

Introduction

A wrapper around go-yaml designed to enable a better way of handling YAML when marshaling to and from structs.

This is a fork and split of the original ghodss/yaml repository which no longer appears to be maintained.

In short, this library first converts YAML to JSON using go-yaml and then uses json.Marshal and json.Unmarshal to convert to or from the struct. This means that it effectively reuses the JSON struct tags as well as the custom JSON methods MarshalJSON and UnmarshalJSON unlike go-yaml. For a detailed overview of the rationale behind this method, see this blog post.

Compatibility

This package uses go-yaml and therefore supports everything go-yaml supports.

Tested against Go versions 1.14 and onwards.

Caveats

Caveat #1: When using yaml.Marshal and yaml.Unmarshal, binary data should NOT be preceded with the !!binary YAML tag. If you do, go-yaml will convert the binary data from base64 to native binary data, which is not compatible with JSON. You can still use binary in your YAML files though - just store them without the !!binary tag and decode the base64 in your code (e.g. in the custom JSON methods MarshalJSON and UnmarshalJSON). This also has the benefit that your YAML and your JSON binary data will be decoded exactly the same way. As an example:

BAD:
	exampleKey: !!binary gIGC

GOOD:
	exampleKey: gIGC
... and decode the base64 data in your code.

Caveat #2: When using YAMLToJSON directly, maps with keys that are maps will result in an error since this is not supported by JSON. This error will occur in Unmarshal as well since you can't unmarshal map keys anyways since struct fields can't be keys.

Installation and usage

To install, run:

$ go get github.com/invopop/yaml

And import using:

import "github.com/invopop/yaml"

Usage is very similar to the JSON library:

package main

import (
	"fmt"

	"github.com/invopop/yaml"
)

type Person struct {
	Name string `json:"name"` // Affects YAML field names too.
	Age  int    `json:"age"`
}

func main() {
	// Marshal a Person struct to YAML.
	p := Person{"John", 30}
	y, err := yaml.Marshal(p)
	if err != nil {
		fmt.Printf("err: %v\n", err)
		return
	}
	fmt.Println(string(y))
	/* Output:
	age: 30
	name: John
	*/

	// Unmarshal the YAML back into a Person struct.
	var p2 Person
	err = yaml.Unmarshal(y, &p2)
	if err != nil {
		fmt.Printf("err: %v\n", err)
		return
	}
	fmt.Println(p2)
	/* Output:
	{John 30}
	*/
}

yaml.YAMLToJSON and yaml.JSONToYAML methods are also available:

package main

import (
	"fmt"

	"github.com/invopop/yaml"
)

func main() {
	j := []byte(`{"name": "John", "age": 30}`)
	y, err := yaml.JSONToYAML(j)
	if err != nil {
		fmt.Printf("err: %v\n", err)
		return
	}
	fmt.Println(string(y))
	/* Output:
	name: John
	age: 30
	*/
	j2, err := yaml.YAMLToJSON(y)
	if err != nil {
		fmt.Printf("err: %v\n", err)
		return
	}
	fmt.Println(string(j2))
	/* Output:
	{"age":30,"name":"John"}
	*/
}