Skip to content

Gradle plugin for simple testing of JSON/XML/SOAP/etc APIs.

License

Notifications You must be signed in to change notification settings

SmartsquareGmbH/squit

Repository files navigation

Squit CI Gradle Plugin Portal

Squit is a Gradle plugin for file-based, automated testing of JSON, XML, SOAP and other apis.
It features high customizability and speed.

Table of Contents

Integration

Add the plugin to your buildscript:

plugins {
    id "de.smartsquare.squit" version "5.1.0"
}

The minimum supported Gradle version is 6.8.

Project structure

Projects are structured in arbitrarily deep folders. The plugin expects the root to be in the src/squit folder per default.

A single test is represented by one leaf folder. That folder must contain:

  • A response file (the file ending depends on the type of test).

Further it can contain:

  • A test.conf file.
  • A description.md file.
  • A request file (the file ending depends on the type of test).
  • db_$name_pre.sql files.
  • db_$name_post.sql files.

The request file contains whatever payload you want to send to your backend. The response file contains the expected response.

A test.conf file is required at least once on the path of your test. That means that it is resolved recursively, starting at the leaf, e.g. your test folder. The test.conf can and must contain various properties, which are discussed in the Configuration section. These properties are then merged if not existing while going up the folder tree.
This allows for convenient definition of properties for multiple tests, with the ability to override properties in special cases.

The description.md file is an optional file containing additional descriptions for tests in the Markdown format. If the tests are nested inside each other and there are multiple description files on the path, they are merged together from top to bottom.

A simple example looks like this:

- src
--- squit
----- my_suite
------- test1 (folder)
--------- request.xml
--------- response.xml
--------- test.conf
--------- description.md
------- test2 (folder)
--------- request.xml
--------- response.xml
------- test.conf

This shows a valid project structure for Squit. my_suite contains all our tests (in this case only two: test1 and test2).

You can have more directories beneath my_suite (e.g. another_suite) and as aforementioned can also nest more deeply. At least one suite folder is required though, you can't have your tests directly in the src/squit folder.

my_suite also contains a test.conf file, which could look like this:

endpoint = "http://localhost:1234/endpoint"

Squit would then use http://localhost:1234/endpoint as the endpoint to call when running all tests in my_suite .
As the example shows, test1 also contains a test.conf file. This one could be used to override the endpoint property of the test.conf file in the my_suite folder.

It is not required to have a test.conf file there, often it is enough to have one for all your tests in the root folder.

Usage

The plugin is composed of various Gradle tasks. For daily usage, only the squitTest is relevant.
The following table lists all tasks and their purpose:

Task name Description
squitPreProcess Pre processes the test sources in a configurable manner.
squitRunRequests Runs the actual requests against your backend.
squitPostProcess Post processes the responses in a configurable manner.
squitTest Compares the expected and actual response and fails the build if differences were found. Also generates the report.

To run all your tests, execute ./gradlew squitTest.

You do NOT need to run the clean task every time you rerun the tests. Squit and Gradle will always run the tests, but only pre-process again if files have actually changed, saving execution time.

Configuration

The plugin features a variety of configuration possibilities. As aforementioned, these are collected in test.conf ( or local.conf) files. test.conf files are in the HOCON format and support all of its features. As of the current version, these are the supported properties:

Name Description Example
endpoint The endpoint of your backend to call. endpoint = "http://localhost:1234/api"
mediaType The media type of your content to send. mediaType = "application/soap+xml"
method The method for the request to use. The default is POST and requires a request.xml. Methods like GET do not require one. method = "GET"
exclude Excludes or un-excludes the test or test group. exclude = true
ignore Ignores or un-ignores the test or test group. This means that the test is run, but does not show up in anything generated at the end of the build. ignore = true
databaseConfigurations An array of database configurations to use for pre- and post scripts. See below for details. databaseConfigurations = [ { / *content */ } ]
preProcessors An array of pre processor classes to use. preProcessors = ["com.example.ExamplePreProcessor"]
postProcessors An array of post processor classes to use. postProcessors = ["com.example.ExamplePostProcessor"]
preProcessorScripts An array of paths to groovy pre processor scripts to use. preProcessorScripts = [./scripts/pre_processor.groovy]
postProcessorScripts An array of paths to groovy post processor scripts to use. postProcessorScripts = [./scripts/post_processor.groovy]
preRunners An array of pre runner classes to use. preRunners= ["com.example.ExamplePreRunner"]
preTestTasks An array of tasks to execute before the test. The order of the entries determines the order of execution. preTestTasks = ["DATABASE_SCRIPTS", "PRE_RUNNERS", "PRE_RUNNER_SCRIPTS"]
postRunners An array of post runner classes to use. postRunners = ["com.example.ExamplePostRunner"]
preRunnerScripts An array of paths to groovy pre runner scripts to use. preRunnerScripts = [./scripts/pre_runner.groovy]
postRunnerScripts An array of paths to groovy post runner scripts to use. postRunnerScripts = [./scripts/post_runner.groovy]
postTestTasks An array of tasks to execute after the test. The order of the entries determines the order of execution. postTestTasks = ["DATABASE_SCRIPTS", "POST_RUNNERS", "POST_RUNNER_SCRIPTS"]
headers A map of headers to use for requests. headers = { "some-header": "value" }
title An optional alternative title for the test. title = "Something"
expectedResponseCode An optional expected HTTP response code. Default is the 200-range. expectedResponseCode = 400

The parameter endpoint is required and the build will fail if it is missing for a test.

Templating

It may be useful to have a placeholder in a test.conf file and fill it at runtime, for example when the port of an endpoint is dynamic or when running in a CI environment.

The HOCON config format which Squit uses comes with support out of the box for it:

endpoint = "http://localhost:"${port}"/someEndpoint"

port could then be replaced when invoking Squit like this:

./gradlew squitTest -Psquit.port=1234

This mechanism can also be used to create global configuration properties, which are then used in configuration files deeper in the hierarchy.

Local configuration

Squit also allows for configuration to be stored in a local.conf file. local.conf files have a higher priority than test.conf file and thus override test.conf files. This can be useful for overriding values of a versioned test.conf without having to check that change into a VCS for every collaborator on the project.

Database modifications

As part of your tests, you may want to modify your database into a specific state. Squit allows you to do so with ordinary sql scripts which can be run before and after a test.

To do so, you have to add a database configuration to your test.conf file(s) and specify the jdbc driver to use.

A simple example would look like this:

# test.conf

databaseConfigurations = [
    {name = "mydb", jdbc = "jdbc:oracle:thin:@localhost:1521:xe", username = "someusername", password = "thepassword"}
    // More are possible.
]
// build.gradle

squit {
    jdbcDrivers = ['oracle.jdbc.driver.OracleDriver'] // You can add more if needed.
}

As you can see, the database properties follow a specific naming scheme. To be recognized, your database configurations must start with db_ and you need all three shown variants (_jdbc, _username, _password). The name in the middle can be arbitrarily chosen and is later used to find the sql scripts.

You would then have to add the jdbc driver to your classpath. This can be done like this (Assuming the jar is in the libs directory of your project):

buildscript {
    dependencies {
        classpath files("libs/ojdbc6.jar")
    }
}

The sql files are added per test. They are required to be named after the configuration you added in the test.conf file, ending with either _pre.sql or _post.sql. The example from before would be mydb_pre.sql and mydb_post.sql.

You can also add a sql script in a higher level of your project structure to merge it into existing scripts. _pre.sql scripts are prepended and _post.sql scrips are appended.

The last option is to have scripts which are only run once. For this, you name the script example_pre_once.sql or example_post_once.sql.

Pre- and Post-processing

Squit allows you to pre- and post-process the requests and actual responses. This may be required for incremental ids you have no control over, dates or other things.

There are currently two ways to do so: Using Groovy scripts or implementing a specific interface.

Groovy processing

Groovy processing is the more easy, but less flexible option.

You add a .groovy script somewhere in your project and supply Squit with the path:

# test.conf

preProcessorScripts = [./some/path/pre_process.groovy]

As for the pre process step, the script gets passed request, expectedResponse and config objects, which types depend on the request type. See supported request types.

A simple script could look like this:

import java.time.LocalDate

request.selectNodes("//Date").each {
    it.text = LocalDate.now().toString()
}

The passed objects for the post-processor are called actualResponse and expectedResponse. Note that the expectedResponse is only for reference here and changes to it are not reflected.

Interface processing

Implementing the Squit interfaces is harder to set up, but much more flexible than the scripting approach.

To do so, you need to set up a buildSrc project or an external project, which is added to the classpath. In the following example, a buildSrc project is set up:

Create the buildSrc folder and set up a normal project in your preferred JVM language. After that, you add the Squit library to your dependencies:

repositories {
    gradlePluginPortal()
    mavenCentral()
}

dependencies {
    compile 'de.smartsquare:squit:5.1.0'
}

Then you can implement one of the interfaces. An example for the SquitXmlPreProcessor equivalent to the scripting example could look like this:

import com.typesafe.config.Config;
import de.smartsquare.squit.SquitPreProcessor;
import org.dom4j.Document;

import java.time.LocalDate;

public class MyPreProcessor implements SquitXmlPreProcessor {

    @Override
    public void process(Document request, Document expectedResponse, Config config) {
        request.selectNodes("//Date")
            .forEach(it -> it.setText(LocalDate.now().toString()));
    }
}

The other interfaces as of the current version are SquitXmlPostProcessor, SquitJsonPostProcessor, SquitJsonPreProcessor.

The last step is to add the class to your test.conf, similar to the approach with groovy scripts:

# test.conf

preProcessors = ["com.example.MyPreProcessor"]

Pre- and Post-runners

Similar to Pre- and Post-processing, it is possible to specify implementations or scripts, which can run arbitrary code before and after each request. The setup is analogous to pre- and post-processors. See the Configuration section for the different options to enable them in your test.conf.

Tagging

Tags allow to run only a subset of your tests to save time and resources if needed.

You can specify tags in corresponding test.conf files. An example could look like this:

tags = ["fast", "mysuite"]

All tests covered by this test.conf file would then be tagged as fast and mysuite.
To run only tests with the tag fast, you would invoke Squit like so:

./gradlew squitTest -Ptags=fast

You can also specify more tags by separating with a ,. Tags are then linked like an "and". If you specify -Ptags=fast,mysuite a test would need to have both tags to be included. If you want to have the semantics of an "or", use -PtagsOr.

Squit also automatically tags your tests named on the folders they reside in. If you have a test in the folder test1 , it would have the tag test1 and could be run exclusively by invoking ./gradlew squitTest -Ptags=test1.

Supported request types

As of the current version, Squit supports these request formats:

Media Type File ending Pre- and post-processor input
application/xml .xml Dom4J Documents
application/json .json Gson JsonElements
All others .txt

Command line options

Because Squit is a Gradle plugin, it supports all of Gradle's command line options and configuration options. Additionally, these options are evaluated:

Option Description Example
tags The tags tests must have to be included in a run. ./gradlew squitTest -Ptags=fast
tagsOr The tags tests must have at least one of to be included in a run. ./gradlew squitTest -Ptags=fast,suite
unignore Handle all tests as if the ignore property would not be set. ./gradlew squitTest -Punignore
unexclude Handle all tests as if the exclude property would not be set. ./gradlew squitTest -Punexclude

Squit Dsl

Here is a complete example of the Squit dsl:

squit {
    // The jdbc drivers to use. Must be on the classpath.
    jdbcDrivers = ['oracle.jdbc.driver.OracleDriver']

    // The path of your test sources. src/squit is the default.
    sourceDir "src/squit"

    // The path to save reports in. build/squit/reports is the default.
    reportDir "build/squit/reports"

    // The timeout in seconds for requests before squit fails. The default is 10.
    timeout = 60

    // If Squit should only print if a tests fails.
    silent = false

    // If the task should pass on test failures or not.
    ignoreFailures = false

    xml {
        // If the xml diffing should be strict.
        // If not, differences like namespace prefixes are not reported as differences.
        strict = true

        // If xml should be canonicalized for the html report.
        // This means that both expected and actual response are transposed into a common format.
        canonicalize = true

        // If invalid namespace resources given in the xml should be resolved.
        // If neither http:// or https:// are set as part of the namespace url, http:// will be set as a prefix,
        // so that the canonicalization does not throw errors.
        resolveInvalidNamespaces = true
    }

    json {
        // If json should be canonicalized for the html report.
        // This means that both expected and actual response are transposed into a common format.
        canonicalize = true
    }
}

All of the shown settings are optional.