Jest-like snapshot testing in Go
- Installation
- MatchSnapshot
- MatchStandaloneSnapshot
New
- MatchJSON
- MatchStandaloneJSON
New
- MatchYAML
New
- Matchers
- Configuration
- Update Snapshots
- Running tests on CI
- No Color
- Snapshots Structure
- Acknowledgments
- Contributing
- Appendix
To install go-snaps
, use go get
:
go get github.com/gkampitakis/go-snaps
Import the go-snaps/snaps
package into your code:
package example
import (
"testing"
"github.com/gkampitakis/go-snaps/snaps"
)
func TestExample(t *testing.T) {
snaps.MatchSnapshot(t, "Hello World")
}
MatchSnapshot
can be used to capture any type of data structured or unstructured.
You can pass multiple parameters to MatchSnapshot
or call MatchSnapshot
multiple
times inside the same test. The difference is in the latter, it will
create multiple entries in the snapshot file.
// test_simple.go
func TestSimple(t *testing.T) {
t.Run("should make multiple entries in snapshot", func(t *testing.T) {
snaps.MatchSnapshot(t, 5, 10, 20, 25)
snaps.MatchSnapshot(t, "some value")
})
}
go-snaps
saves the snapshots in __snapshots__
directory and the file
name is the test file name with extension .snap
.
So for example if your test is called test_simple.go
when you run your tests, a snapshot file
will be created at ./__snapshots__/test_simple.snaps
.
MatchStandaloneSnapshot
will create snapshots on separate files as opposed to MatchSnapshot
which adds multiple snapshots inside the same file.
Combined with snaps.Ext
you can have proper syntax highlighting and better readability
// test_simple.go
func TestSimple(t *testing.T) {
snaps.MatchStandaloneSnapshot(t, "Hello World")
// or create an html snapshot file
snaps.WithConfig(snaps.Ext(".html")).
MatchStandaloneSnapshot(t, "<html><body><h1>Hello World</h1></body></html>")
}
go-snaps
saves the snapshots in __snapshots__
directory and the file
name is the t.Name()
plus a number plus the extension .snap
.
So for the above example the snapshot file name will be ./__snapshots__/TestSimple_1.snap
and ./__snapshots__/TestSimple_1.snap.html
.
MatchJSON
can be used to capture data that can represent a valid json.
You can pass a valid json in form of string
or []byte
or whatever value can be passed
successfully on json.Marshal
.
func TestJSON(t *testing.T) {
type User struct {
Age int
Email string
}
snaps.MatchJSON(t, `{"user":"mock-user","age":10,"email":"mock@email.com"}`)
snaps.MatchJSON(t, []byte(`{"user":"mock-user","age":10,"email":"mock@email.com"}`))
snaps.MatchJSON(t, User{10, "mock-email"})
}
JSON will be saved in snapshot in pretty format for more readability and deterministic diffs.
MatchStandaloneJSON
will create snapshots on separate files as opposed to MatchJSON
which adds multiple snapshots inside the same file.
func TestSimple(t *testing.T) {
snaps.MatchStandaloneJSON(t, `{"user":"mock-user","age":10,"email":"mock@email.com"}`)
snaps.MatchStandaloneJSON(t, User{10, "mock-email"})
}
go-snaps
saves the snapshots in __snapshots__
directory and the file
name is the t.Name()
plus a number plus the extension .snap.json
.
So for the above example the snapshot file name will be ./__snapshots__/TestSimple_1.snap.json
and ./__snapshots__/TestSimple_2.snap.json
.
MatchYAML
can be used to capture data that can represent a valid yaml.
You can pass a valid json in form of string
or []byte
or whatever value can be passed
successfully on yaml.Marshal
.
func TestYAML(t *testing.T) {
type User struct {
Age int
Email string
}
snaps.MatchYAML(t, "user: \"mock-user\"\nage: 10\nemail: mock@email.com")
snaps.MatchYAML(t, []byte("user: \"mock-user\"\nage: 10\nemail: mock@email.com"))
snaps.MatchYAML(t, User{10, "mock-email"})
}
MatchJSON
's and MatchYAML
's third argument can accept a list of matchers. Matchers are functions that can act
as property matchers and test values.
You can pass the path of the property you want to match and test.
Currently go-snaps
has three build in matchers
match.Any
match.Custom
match.Type[ExpectedType]
Open to feedback for building more matchers or you can build your own example.
For JSON go-snaps utilises gjson.
More information about the supported path syntax from gjson.
As for YAML go-snaps utilises github.com/goccy/go-yaml#5-use-yamlpath.
More information about the supported syntax PathString.
Any matcher acts as a placeholder for any value. It replaces any targeted path with a placeholder string.
Any("user.name")
// or with multiple paths
Any("user.name", "user.email")
Any matcher provides some methods for setting options
match.Any("user.name").
Placeholder(value). // allows to define a different placeholder value from the default "<Any Value>"
ErrOnMissingPath(bool) // determines whether the matcher will err in case of a missing, default true
Custom matcher allows you to bring your own validation and placeholder value
match.Custom("user.age", func(val any) (any, error) {
age, ok := val.(float64)
if !ok {
return nil, fmt.Errorf("expected number but got %T", val)
}
return "some number", nil
})
The callback parameter value for JSON can be on of these types:
bool // for JSON booleans
float64 // for JSON numbers
string // for JSON string literals
nil // for JSON null
map[string]any // for JSON objects
[]any // for JSON arrays
If Custom matcher returns an error the snapshot test will fail with that error.
Custom matcher provides a method for setting an option
match.Custom("path",myFunc).
Placeholder(value). // allows to define a different placeholder value from the default "<Any Value>"
ErrOnMissingPath(bool) // determines whether the matcher will err in case of a missing path, default true
Type matcher evaluates types that are passed in a snapshot and it replaces any targeted path with a placeholder in the form of <Type:ExpectedType>
.
match.Type[string]("user.info")
// or with multiple paths
match.Type[float64]("user.age", "data.items")
Type matcher provides a method for setting an option
match.Type[string]("user.info").
ErrOnMissingPath(bool) // determines whether the matcher will err in case of a missing path, default true
You can see more examples.
go-snaps
allows passing configuration for overriding
- the directory where snapshots are stored, relative or absolute path
- the filename where snapshots are stored
- the snapshot file's extension (regardless the extension the filename will include the
.snaps
inside the filename) - programmatically control whether to update snapshots. You can find an example usage at examples
t.Run("snapshot tests", func(t *testing.T) {
snaps.WithConfig(snaps.Filename("my_custom_name"), snaps.Dir("my_dir")).MatchSnapshot(t, "Hello Word")
s := snaps.WithConfig(
snaps.Dir("my_dir"),
snaps.Filename("json_file"),
snaps.Ext(".json")
snaps.Update(false)
)
s.MatchJSON(t, `{"hello":"world"}`)
})
You can see more on examples
You can update your failing snapshots by setting UPDATE_SNAPS
env variable to true.
UPDATE_SNAPS=true go test ./...
If you don't want to update all failing snapshots, or you want to update only one of
them you can you use the -run
flag to target the test(s) you want.
For more information on go test
flags you can run
go help testflag
go-snaps
can identify obsolete snapshots.
In order to enable this functionality you need to use TestMain(m *testing.M)
to
call snaps.Clean(t)
after your tests have run. This will also print a Snapshot Summary. (if running tests
with verbose flag -v
)
If you want to remove the obsolete snap files and snapshots you can run
tests with UPDATE_SNAPS=clean
env variable.
The reason for using TestMain
is because go-snaps
needs to be sure that all tests
are finished so it can keep track of which snapshots were not called.
Example:
func TestMain(m *testing.M) {
v := m.Run()
// After all tests have run `go-snaps` can check for unused snapshots
snaps.Clean(m)
os.Exit(v)
}
For more information around TestMain.
By default go-snaps
appends new snaps to the snapshot file and in case of parallel tests the order is random. If you want snaps to be sorted in deterministic order you need to use TestMain
per package:
func TestMain(m *testing.M) {
v := m.Run()
// After all tests have run `go-snaps` will sort snapshots
snaps.Clean(m, snaps.CleanOpts{Sort: true})
os.Exit(v)
}
If you want to skip one test using t.Skip
, go-snaps
can't keep track
if the test was skipped or if it was removed. For that reason go-snaps
exposes
a wrapper for t.Skip
, t.Skipf
and t.SkipNow
, which keep tracks of skipped files.
You can skip, or only run specific tests by using the -run
flag. go-snaps
can identify which tests are being skipped and parse only the relevant tests
for obsolete snapshots.
When go-snaps
detects that it is running in CI it will automatically fail when snapshots are missing. This is done to ensure new snapshots are committed alongside the tests and assertions are successful.
go-snaps
uses ciinfo for detecting if it runs on CI environment.
go-snaps
supports disabling color outputs by running your tests with the env variable
NO_COLOR
set to any value.
NO_COLOR=true go test ./...
For more information around NO_COLOR.
Snapshots have the form
[TestName - Number]
<data>
---
TestID
is the test name plus an increasing number to allow multiple calls of MatchSnapshot
in a single test.
[TestSimple/should_make_a_map_snapshot - 1]
map[string]interface{}{
"mock-0": "value",
"mock-1": int(2),
"mock-2": func() {...},
"mock-3": float32(10.399999618530273),
}
---
Note
If your snapshot data contain characters ---
at the start of a line followed by a new line, go-snaps
will "escape" them and save them as /-/-/-/
to differentiate them from termination characters.
This library used Jest Snapshoting and Cupaloy as inspiration.
- Jest is a full-fledged Javascript testing framework and has robust snapshoting features.
- Cupaloy is a great and simple Golang snapshoting solution.
- The logo was made by MariaLetta.
Warning
When running a specific test file by specifying a path go test ./my_test.go
, go-snaps
can't track the path so it will mistakenly mark snapshots as obsolete.
Important
Snapshots should be treated as code. The snapshot artifact should be committed alongside code changes, and reviewed as part of your code review process
Note
go-snaps doesn't handle CRLF line endings. If you are using Windows, you may need to convert the line endings to LF.