Litter is a pretty printer library for Go data structures to aid in debugging and testing.
Litter is provided by
Sanity: The Headless CMS Construction Kit
Litter named for the fact that it outputs literals, which you litter your output with. As a side benefit, all Litter output is syntactically correct Go. You can use Litter to emit data during debug, and it's also really nice for "snapshot data" in unit tests, since it produces consistent, sorted output. Litter was inspired by Spew, but focuses on terseness and readability.
This:
type Person struct {
Name string
Age int
Parent *Person
}
litter.Dump(Person{
Name: "Bob",
Age: 20,
Parent: &Person{
Name: "Jane",
Age: 50,
},
})
will output:
Person{
Name: "Bob",
Age: 20,
Parent: &Person{
Name: "Jane",
Age: 50,
},
}
Litter is a great alternative to JSON or YAML for providing "snapshots" or example data. For example:
func TestSearch(t *testing.T) {
result := DoSearch()
actual := litterOpts.Sdump(result)
expected, err := ioutil.ReadFile("testdata.txt")
if err != nil {
// First run, write test data since it doesn't exist
if !os.IsNotExist(err) {
t.Error(err)
}
ioutil.Write("testdata.txt", actual, 0644)
actual = expected
}
if expected != actual {
t.Errorf("Expected %s, got %s", expected, actual)
}
}
The first run will use Litter to write the data to testdata.txt
. On subsequent runs, the test will compare the data. Since Litter always provides a consistent view of a value, you can compare the strings directly.
Litter detects circular references or aliasing, and will replace additional references to the same object with aliases. For example:
type Circular struct {
Self *Circular
}
selfref := Circular{}
selfref.Self = &selfref
litter.Dump(selfref)
will output:
Circular { // p0
Self: p0,
}
$ go get -u github.com/sanity-io/litter
Add this import line to the file you're working in:
import "github.com/sanity-io/litter"
To dump a variable with full newlines, indentation, type, and aliasing information, use Dump
or Sdump
:
litter.Dump(myVar1)
str := litter.Sdump(myVar1)
Dumps the data structure to STDOUT.
Returns the dump as a string
You can configure litter globally by modifying the default litter.Config
// Strip all package names from types
litter.Config.StripPackageNames = true
// Hide private struct fields from dumped structs
litter.Config.HidePrivateFields = true
// Hide fields matched with given regexp if it is not nil. It is set up to hide fields generate with protoc-gen-go
litter.Config.FieldExclusions = regexp.MustCompile(`^(XXX_.*)$`)
// Sets a "home" package. The package name will be stripped from all its types
litter.Config.HomePackage = "mypackage"
// Sets separator used when multiple arguments are passed to Dump() or Sdump().
litter.Config.Separator = "\n"
// Use compact output: strip newlines and other unnecessary whitespace
litter.Config.Compact = true
// Prevents duplicate pointers from being replaced by placeholder variable names (except in necessary, in the case
// of circular references)
litter.Config.DisablePointerReplacement = true
Allows you to configure a local configuration of litter to allow for proper compartmentalization of state at the expense of some comfort:
sq := litter.Options {
HidePrivateFields: true,
HomePackage: "thispack",
Separator: " ",
}
sq.Dump("dumped", "with", "local", "settings")
Implement the interface Dumper on your types to take control of how your type is dumped.
type Dumper interface {
LitterDump(w io.Writer)
}
Just write your custom dump to the provided stream, using multiple lines divided by "\n"
if you need. Litter
might indent your output according to context, and optionally decorate your first line with a pointer comment
where appropriate.
A couple of examples from the test suite:
type CustomMultiLineDumper struct {}
func (cmld *CustomMultiLineDumper) LitterDump(w io.Writer) {
w.Write([]byte("{\n multi\n line\n}"))
}
type CustomSingleLineDumper int
func (csld CustomSingleLineDumper) LitterDump(w io.Writer) {
w.Write([]byte("<custom>"))
}