- Inspired by facebook's Jest framework
🎉 4.0.0 is out
- Upgrade guide from 3.X to 4.X here
- Upgrade guide from 2.X to 3.X here
- Upgrade guide from 2.X-BETA to 2.X here
- Tired of needing to
assertThat(foo).isEqualTo("bar")
again & again? - Are you just wanting to ensure you don't break - for example - REST interfaces
- Are you manually saving text files for verification in your tests?
Want a better way? Then java-snapshot-testing might just be what you are looking for!
- Add test dependencies
// In this case we are using the JUnit5 testing framework
testImplementation 'io.github.origin-energy:java-snapshot-testing-junit5:4.+'
// slf4j logging implementation if you don't already have one
testImplementation("org.slf4j:slf4j-simple:2.0.0-alpha0")
// Optional: Many will want to serialize into JSON. In this case you should also add the Jackson plugin
testImplementation 'io.github.origin-energy:java-snapshot-testing-plugin-jackson:4.+'
testImplementation 'com.fasterxml.jackson.core:jackson-core:2.11.3'
testImplementation 'com.fasterxml.jackson.core:jackson-databind:2.11.3'
// Optional: If you want Jackson to serialize Java 8 date/time types or Optionals you should also add the following dependencies
testRuntimeOnly 'com.fasterxml.jackson.datatype:jackson-datatype-jdk8:2.11.3'
testRuntimeOnly 'com.fasterxml.jackson.datatype:jackson-datatype-jsr310:2.11.3'
- Create
snapshot.properties
and configure your global settings. Be sure to setoutput-dir
appropriately for your JVM language.
- /src/test/resources/snapshot.properties
serializer=au.com.origin.snapshots.serializers.v1.ToStringSnapshotSerializer
serializer.base64=au.com.origin.snapshots.serializers.v1.Base64SnapshotSerializer
serializer.json=au.com.origin.snapshots.jackson.serializers.v1.JacksonSnapshotSerializer
serializer.orderedJson=au.com.origin.snapshots.jackson.serializers.v1.DeterministicJacksonSnapshotSerializer
comparator=au.com.origin.snapshots.comparators.v1.PlainTextEqualsComparator
reporters=au.com.origin.snapshots.reporters.v1.PlainTextSnapshotReporter
snapshot-dir=__snapshots__
output-dir=src/test/java
ci-env-var=CI
update-snapshot=none
- Enable snapshot testing and write your first test
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Expect;
import au.com.origin.snapshots.annotations.SnapshotName;
import au.com.origin.snapshots.junit5.SnapshotExtension;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.ExtendWith;
import java.util.HashMap;
import java.util.Map;
@ExtendWith({SnapshotExtension.class})
public class MyFirstSnapshotTest {
private Expect expect;
@SnapshotName("i_can_give_custom_names_to_my_snapshots")
@Test
public void toStringSerializationTest() {
expect.toMatchSnapshot("Hello World");
}
@Test
public void jsonSerializationTest() {
Map<String, Object> map = new HashMap<>();
map.put("name", "John Doe");
map.put("age", 40);
expect
.serializer("json")
.toMatchSnapshot(map);
}
}
- Run your test
Bingo - you should now see your snapshot in the __snapshots__
folder created next to your test. Try
changing "Hello World"
to "Hello Universe"
and watch it fail with a .debug
file.
au.com.origin.snapshots.docs.MyFirstSnapshotTest.jsonSerializationTest=[
{
"age": 40,
"name": "John Doe"
}
]
i_can_give_custom_names_to_my_snapshots=[
Hello World
]
- Great for testing JSON interfaces ensuring you don't break clients
- Fast and easy to test
- Will implicitly test areas of your code you did not think about
- Great of testing dynamic objects
You're responsible for making sure your generated snapshots do not include platform specific or other non-deterministic data.
- You need to ensure your test is deterministic for all fields (there are ways to ignore things like dates)
- Does not give great insight to why the snapshot failed
- Can be difficult to troll though large snapshot changes where you might only be interested in a small set of fields
Installation Maven
These docs are for the latest -SNAPSHOT
version published to maven central. Select the tag X.X.X
matching your maven
dependency to get correct documentation for your version.
Only if you want to integrate with an unsupported framework. Show me how!
We currently support:
Plugins
- Jackson for JSON serialization
- You need jackson on your classpath (Gradle example)
// Required java-snapshot-testing peer dependencies testImplementation 'com.fasterxml.jackson.core:jackson-core:2.11.3' testImplementation 'com.fasterxml.jackson.core:jackson-databind:2.11.3' // Optional java-snapshot-testing peer dependencies testRuntimeOnly 'com.fasterxml.jackson.datatype:jackson-datatype-jdk8:2.11.3' testRuntimeOnly 'com.fasterxml.jackson.datatype:jackson-datatype-jsr310:2.11.3'
- You need jackson on your classpath (Gradle example)
- When a test runs for the first time, a
.snap
file is created in a__snapshots__
sub-directory - On subsequent test runs, the
.snap
file is compared with the one produced by the test - If they don't match, the test fails and a
.snap.debug
with the conflict is created - It is then your job to decide if you have introduced a regression or intentionally changed the output (Use your IDE file comparison tools to compare the two files or refer to the terminal output)
- If you have introduced a regression you will need to fix your code
- If you have intentionally changed the output you can manually modify the
.snap
file to make it pass or delete it and it will be generated again from scratch - Once you fix the test, the
*.snap.debug
file will get deleted
A text representation of your java object (toString() or JSON).
String snapshot example
expect.toMatchSnapshot("Hello World");
au.com.example.company.HelloWorldTest.helloWorld=[
Hello world
]
JSON Snapshot Example
expect.serializer("json").toMatchSnapshot(userDto);
au.com.example.company.UserEndpointTest.shouldReturnCustomerData=[
{
"id": "1",
"firstName": "John",
"lastName": "Smith",
"age": 34
}
]
All frameworks allow injection of the Expect expect
via instance variable or method argument. In cases where
parameterised tests are used, it's often better to use an instance variable in order to avoid conflicts with
the underlying data table.
Note: Due to the above restriction, method argument injection is destined for removal in future versions.
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.junit5.SnapshotExtension;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.ExtendWith;
import au.com.origin.snapshots.Expect;
// Ensure you extend your test class with the SnapshotExtension
@ExtendWith({SnapshotExtension.class})
public class JUnit5Example {
// Option 1: inject Expect as an instance variable
private Expect expect;
@Test
public void myTest1() {
// Verify your snapshot
expect.toMatchSnapshot("Hello World");
}
// Option 2: inject Expect into the method signature
@Test
public void myTest2(Expect expect) {
expect.toMatchSnapshot("Hello World Again");
}
}
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.annotations.SnapshotName;
import au.com.origin.snapshots.junit4.SnapshotRunner;
import au.com.origin.snapshots.Expect;
import org.junit.Test;
import org.junit.runner.RunWith;
// Ensure you RunWith the SnapshotRunner
@RunWith(SnapshotRunner.class)
public class JUnit4Example {
// Option 1: inject Expect as an instance variable
private Expect expect;
@SnapshotName("my first test")
@Test
public void myTest1() {
// Verify your snapshot
expect.toMatchSnapshot("Hello World");
}
@SnapshotName("my second test")
@Test
// Option 2: inject Expect into the method signature
public void myTest2(Expect expect) {
expect.toMatchSnapshot("Hello World Again");
}
}
In order to run alongside another JUnit4 test runner such as @RunWith(Parameterized.class)
, you need to use the
Rule based configuration instead.
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Expect;
import au.com.origin.snapshots.annotations.SnapshotName;
import au.com.origin.snapshots.junit4.SnapshotClassRule;
import au.com.origin.snapshots.junit4.SnapshotRule;
import org.junit.ClassRule;
import org.junit.Rule;
import org.junit.Test;
public class JUnit4RulesExample {
@ClassRule
public static SnapshotClassRule snapshotClassRule = new SnapshotClassRule();
@Rule
public SnapshotRule snapshotRule = new SnapshotRule(snapshotClassRule);
private Expect expect;
@SnapshotName("my first test")
@Test
public void myTest1() {
expect.toMatchSnapshot("Hello World");
}
}
See the ParameterizedTest for an example implementation
package au.com.origin.snapshots.docs
import au.com.origin.snapshots.annotations.SnapshotName
import au.com.origin.snapshots.spock.EnableSnapshots
import spock.lang.Specification
import au.com.origin.snapshots.Expect
// Ensure you enable snapshot testing support
@EnableSnapshots
class SpockExample extends Specification {
// Option 1: inject Expect as an instance variable
private Expect expect
// With spock tests you should always use @SnapshotName - otherwise they become coupled to test order
@SnapshotName("should_use_extension")
def "Should use extension"() {
when:
expect.toMatchSnapshot("Hello World")
then:
true
}
@SnapshotName("should_use_extension_as_method_argument")
// Option 2: inject Expect into the method signature
def "Should use extension as method argument"(Expect expect) {
when:
expect.toMatchSnapshot("Hello World")
then:
true
}
}
This library is in no way restricted to JUnit4, Junit5 or Spock.
Any framework can support the library as long as it follows the following rules:
- Before all the tests in a single file execute (once only)
SnapshotVerifier snapshotVerifier = new SnapshotVerifier(new YourFrameworkSnapshotConfig(), testClass, failOnOrphans);
- After all the tests in a single file execute (once only)
snapshotVerifier.validateSnapshots();
- For each test class, setup your expectations
Expect expect = Expect.of(snapshotVerifier, testMethod); expect.toMatchSnapshot("Something");
Here is a JUnit5 example that does not use the JUnit5 extension
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Expect;
import au.com.origin.snapshots.SnapshotVerifier;
import au.com.origin.snapshots.config.PropertyResolvingSnapshotConfig;
import org.junit.jupiter.api.AfterAll;
import org.junit.jupiter.api.BeforeAll;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.TestInfo;
// Notice we aren't using any framework extensions
public class CustomFrameworkExample {
private static SnapshotVerifier snapshotVerifier;
@BeforeAll
static void beforeAll() {
snapshotVerifier = new SnapshotVerifier(new PropertyResolvingSnapshotConfig(), CustomFrameworkExample.class);
}
@AfterAll
static void afterAll() {
snapshotVerifier.validateSnapshots();
}
@Test
void shouldMatchSnapshotOne(TestInfo testInfo) {
Expect expect = Expect.of(snapshotVerifier, testInfo.getTestMethod().get());
expect.toMatchSnapshot("Hello World");
}
}
By default, snapshots use the full method name as the identifier For example
au.com.origin.snapshots.docs.MyFirstSnapshotTest.helloWorldTest=[
Hello World
]
This strategy has a number of problems
- it's long and unwieldy
- if the method name or class name changes, your tests become orphans
- The Spock framework tests use a generated method name that is based on index (Spock tests should always use
@SnapshotName
)
You can supply a more meaningful name to your snapshot using @SnapshotName("your_custom_name")
This will generate as follows
your_custom_name=[
Hello World
]
Much more concise and not affected by class name or method name refactoring.
Often your IDE has an excellent file comparison tool.
- A
*.snap.debug
file will be created alongside your*.snap
file when a conflict occurs. - You can then use your IDE tooling to compare the two files.
*.snap.debug
is deleted automatically once the test passes.
Note: *.snap.debug
files should never be checked into version control so consider adding it to your .gitignore
This file allows you to conveniently setup global defaults
key | Description |
---|---|
serializer | Class name of the serializer, default serializer |
serializer.{name} | Class name of the serializer, accessible via .serializer("{name}") |
comparator | Class name of the comparator |
comparator.{name} | Class name of the comparator, accessible via .comparator("{name}") |
reporters | Comma separated list of class names to use as reporters |
reporters.{name} | Comma separated list of class names to use as reporters, accessible via .reporters("{name}") |
snapshot-dir | Name of sub-folder holding your snapshots |
output-dir | Base directory of your test files (although it can be a different directory if you want) |
ci-env-var | Name of environment variable used to detect if we are running on a Build Server |
update-snapshot | Similar to --updateSnapshot in Jest [all]=update all snapshots [none]=update no snapshots [MyTest1,MyTest2]=update snapshots in these classes only |
For example:
serializer=au.com.origin.snapshots.serializers.v1.ToStringSnapshotSerializer
serializer.base64=au.com.origin.snapshots.serializers.v1.Base64SnapshotSerializer
serializer.json=au.com.origin.snapshots.jackson.serializers.v1.JacksonSnapshotSerializer
serializer.orderedJson=au.com.origin.snapshots.jackson.serializers.v1.DeterministicJacksonSnapshotSerializer
comparator=au.com.origin.snapshots.comparators.v1.PlainTextEqualsComparator
reporters=au.com.origin.snapshots.reporters.v1.PlainTextSnapshotReporter
snapshot-dir=__snapshots__
output-dir=src/test/java
ci-env-var=CI
update-snapshot=none
In cases where the same test runs multiple times with different parameters you need to set the scenario
and it must be
unique for each run
expect.scenario(params).toMatchSnapshot("Something");
package au.com.origin.snapshots.docs
import au.com.origin.snapshots.Expect
import au.com.origin.snapshots.annotations.SnapshotName
import au.com.origin.snapshots.spock.EnableSnapshots
import spock.lang.Specification
@EnableSnapshots
class SpockWithParametersExample extends Specification {
private Expect expect
@SnapshotName("convert_to_uppercase")
def 'Convert #scenario to uppercase'() {
when: 'I convert to uppercase'
String result = value.toUpperCase();
then: 'Should convert letters to uppercase'
// Check you snapshot against your output using a unique scenario
expect.scenario(scenario).toMatchSnapshot(result)
where:
scenario | value
'letter' | 'a'
'number' | '1'
}
}
The serializer determines how a class gets converted into a string.
Serializers are pluggable, so you can write you own by implementing the SnapshotSerializer
interface.
Currently, we support the following serializers.
Serializer | Description |
---|---|
ToStringSnapshotSerializer | uses the toString() method |
Base64SnapshotSerializer | use for images or other binary sources that output a byte[] . The output is encoded to Base64 |
Serializer | Description |
---|---|
JacksonSnapshotSerializer | uses jackson to convert a class to a snapshot |
DeterministicJacksonSnapshotSerializer | extension of JacksonSnapshotSerializer that also orders Collections for situations where the order changes on multiple runs |
Serializers are resolved in the following order.
- (method level) explicitly
expect.serializer(ToStringSerializer.class).toMatchSnapshot(...);
or via property fileexpect.serializer("json").toMatchSnapshot(...);
- (class level) explicitly
@UseSnapshotConfig
which gets read from thegetSerializer()
method - (properties) implicitly via
snapshot.properties
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Expect;
import au.com.origin.snapshots.annotations.UseSnapshotConfig;
import au.com.origin.snapshots.junit5.SnapshotExtension;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.ExtendWith;
@ExtendWith(SnapshotExtension.class)
@UseSnapshotConfig(LowercaseToStringSnapshotConfig.class)
public class JUnit5ResolutionHierarchyExample {
private Expect expect;
@Test
public void aliasMethodTest() {
expect
.serializer("json") // <------ Using snapshot.properties
.toMatchSnapshot(new TestObject());
}
@Test
public void customSerializerTest() {
expect
.serializer(UppercaseToStringSerializer.class) // <------ Using custom serializer
.toMatchSnapshot(new TestObject());
}
// Read from LowercaseToStringSnapshotConfig defined on the class
@Test
public void lowercaseTest() {
expect.toMatchSnapshot(new TestObject());
}
}
Sometimes the default serialization doesn't work for you. An example is Hibernate serialization where you get infinite recursion on Lists/Sets.
You can supply any serializer you like Gson, Jackson or something else.
For example, the following will exclude the rendering of Lists without changing the source code to include @JsonIgnore
. This is good because you shouldn't need to add annotations to your source code for testing purposes only.
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.jackson.serializers.DeterministicJacksonSnapshotSerializer;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.annotation.JsonIgnore;
import com.fasterxml.jackson.annotation.JsonIgnoreType;
import com.fasterxml.jackson.databind.ObjectMapper;
import java.time.Instant;
import java.util.List;
import java.util.Set;
public class HibernateSnapshotSerializer extends DeterministicJacksonSnapshotSerializer {
@Override
public void configure(ObjectMapper objectMapper) {
super.configure(objectMapper);
// Ignore Hibernate Lists to prevent infinite recursion
objectMapper.addMixIn(List.class, IgnoreTypeMixin.class);
objectMapper.addMixIn(Set.class, IgnoreTypeMixin.class);
// Ignore Fields that Hibernate generates for us automatically
objectMapper.addMixIn(BaseEntity.class, IgnoreHibernateEntityFields.class);
}
@JsonIgnoreType
class IgnoreTypeMixin {
}
abstract class IgnoreHibernateEntityFields {
@JsonIgnore
abstract Long getId();
@JsonIgnore
abstract Instant getCreatedDate();
@JsonIgnore
abstract Instant getLastModifiedDate();
}
}
The comparator determines if two snapshots match.
Currently, we support one default comparator (PlainTextEqualsComparator
) which uses string equals for comparison.
This should work for most cases. Custom implementations of SnapshotComparator
can provide more advanced comparisons.
Comparators follow the same resolution order as Serializers
- method
- class
- snapshot.properties
The default comparator may be too strict for certain types of data. For example, when comparing json objects, formatting of the json string or the order of fields may not be of much importance during comparison. A custom comparator can help in such cases.
For example, the following will convert a json string to a Map and then perform an equals comparison so that formatting and field order are ignored.
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Snapshot;
import au.com.origin.snapshots.comparators.SnapshotComparator;
import com.fasterxml.jackson.databind.ObjectMapper;
import lombok.SneakyThrows;
public class JsonObjectComparator implements SnapshotComparator {
@Override
public boolean matches(Snapshot previous, Snapshot current) {
return asObject(previous.getName(), previous.getBody()).equals(asObject(current.getName(), current.getBody()));
}
@SneakyThrows
private static Object asObject(String snapshotName, String json) {
return new ObjectMapper().readValue(json.replaceFirst(snapshotName + "=", ""), Object.class);
}
}
The reporter reports the details of comparison failures.
Currently, we support one default reporter (PlainTextSnapshotReporter
) which uses assertj's DiffUtils to generate a
patch of the differences between two snapshots.
Custom reporters can be plugged in by implementing SnapshotReporter
.
Reporters follow the same resolution order as Serializers and Comparators
For generating and reporting json diffs using other libraries like https://github.com/skyscreamer/JSONassert a custom reporter can be created like the one below.
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Snapshot;
import au.com.origin.snapshots.reporters.SnapshotReporter;
import au.com.origin.snapshots.serializers.SerializerType;
import lombok.SneakyThrows;
import org.skyscreamer.jsonassert.JSONAssert;
import org.skyscreamer.jsonassert.JSONCompareMode;
public class JsonAssertReporter implements SnapshotReporter {
@Override
public boolean supportsFormat(String outputFormat) {
return SerializerType.JSON.name().equalsIgnoreCase(outputFormat);
}
@Override
@SneakyThrows
public void report(Snapshot previous, Snapshot current) {
JSONAssert.assertEquals(previous.getBody(), current.getBody(), JSONCompareMode.STRICT);
}
}
You can add metadata to your snapshots via headers. Headers can be used by Serializers, Comparators & Reporters to help interrogate the snapshot.
Custom Serializers can also inject default headers as needed.
Example of injecting a header manually
String obj = "hello"
expect
.header("className", obj.getClass().getName())
.header("foo", "bar")
.toMatchSnapshot(obj);
Snapshot output
au.com.origin.snapshots.SnapshotHeaders.canAddHeaders={
"className": "java.lang.String",
"foo": "bar"
}[
hello
]
You can override the snapshot configuration easily using the @UseSnapshotConfig
annotation
JUnit5 Example
package au.com.origin.snapshots.docs;
import au.com.origin.snapshots.Expect;
import au.com.origin.snapshots.annotations.UseSnapshotConfig;
import au.com.origin.snapshots.junit5.SnapshotExtension;
import org.junit.jupiter.api.Test;
import org.junit.jupiter.api.extension.ExtendWith;
@ExtendWith(SnapshotExtension.class)
@UseSnapshotConfig(LowercaseToStringSnapshotConfig.class)
public class JUnit5ResolutionHierarchyExample {
private Expect expect;
@Test
public void aliasMethodTest() {
expect
.serializer("json") // <------ Using snapshot.properties
.toMatchSnapshot(new TestObject());
}
@Test
public void customSerializerTest() {
expect
.serializer(UppercaseToStringSerializer.class) // <------ Using custom serializer
.toMatchSnapshot(new TestObject());
}
// Read from LowercaseToStringSnapshotConfig defined on the class
@Test
public void lowercaseTest() {
expect.toMatchSnapshot(new TestObject());
}
}
I'm seeing this error in my logs
org/slf4j/LoggerFactory
java.lang.NoClassDefFoundError: org/slf4j/LoggerFactory
Solution:
Add an SLF4J Provider such astestImplementation("org.slf4j:slf4j-simple:2.0.0-alpha0")
My test source files are not in src/test/java
Solution: Override output-dir
in snapshot.properties
I see the following error in JSON snapshots java.lang.NoSuchFieldError: BINARY
Solution: This happened to me in a spring-boot app, I removed my jackson dependencies and relied on the ones from spring-boot instead.
see CONTRIBUTING.md