Groovy: The “Enterprise Hipster” Language
Not everyone sees the Java programming language as sexy. However, the Java virtual machine is a dominant force everywhere, from the most conservative enterprise to the most whimsical startup. There are myriad alternative languages today that compile to Java bytecode. There are JVM-based versions of Python, Ruby, and multiple implementations of JavaScript. There are entirely new languages, such as Kotlin from JetBrains and Ceylon from RedHat. Clojure has recently re-awakened interest in Lisp, and of course Scala is largely responsible for the 2000’s shift toward functional programming on the server-side.
Groovy, the granddaddy of them all, is today almost invisible in its ubiquity. When it first appeared 13 years ago, Groovy was an instant hit. The language, and the related Grails web framework, combined the emerging popularity of Ruby on Rails with an extremely shallow learning curve for Java developers. Virtually overnight, Groovy completely replaced BeanShell, the previous JVM-scripting alternative.
Enthusiasm for the Rails model eventually waned, and strongly-typed languages became the trend once more. Quite frankly, many people who flocked to Groovy simply because it was “new” continued to move on to newer things. However, Groovy didn’t fade away. Rather, it settled into a mature role as the “enterprise hipster” language. You find it everywhere. Almost any application on the JVM that exposes a scripting interface does so with Groovy as the first-class citizen. Groovy is extremely popular with QA in the automated testing space, is deeply baked into the Spring framework, and is the basis for the fast-growing Gradle build system.
We don’t hype Groovy as much as we used to, but it’s absolutely entrenched in the Java ecosystem, and is continuing to expand. It’s a stable, safe bet, for which it’s easy to find talent (or quickly train it on-the-job). While there are more trendy buzzwords to put on your resume today, there seems little risk of Groovy flaming out and going away anytime soon. Groovy “just works”, and is a really convenient tool that every Java developer should have in his or her toolbox.
Gradle as a Groovy App Server
History aside, let’s talk about a recent use case that lead me to dust off my Groovy skills. I needed to quickly stand up a registry of “key-value” config parameters, for a number of applications running within a number of environments. I wanted to capture these parameters in source control as a collection of properties files. One file for each application, nested within a subdirectory for each environment:
- …
- qa-env/
- application-a.properties
- application-b.properties
- …
- staging-env/
- application-a.properties
- application-b.properties
- …
- …
Whenever a change to these properties files is committed in source control, I would like Jenkins (or some other continuous integration server) to sync its values with a runtime “registry”. That registry might eventually become something like etcd, or Consul and Vault, but we can get things started quickly with a traditional MySQL database.
Since most of our continuous integration build jobs are Gradle-based these days, and since Gradle is Groovy-native, we can bake this “sync” job into a Gradle build. With a JavaExec-based task, pointing to a Groovy script, you can leverage Gradle as a Groovy app server!
build.gradle
apply plugin: 'groovy'
repositories {
mavenCentral()
mavenLocal()
}
// [1] Declare a localGroovy() dependency, to use
// the Groovy library that ships with Gradle.
dependencies {
compile localGroovy()
compile("mysql:mysql-connector-java:5.1.35")
compile("com.h2database:h2:1.4.187")
testCompile("junit:junit:4.12")
}
// [2] Create a task of type 'JavaExec', referencing
// a Groovy script and any input arguments.
task runScript(type: JavaExec) {
description 'Run a Groovy script to sync the environment config registry with the properties files in source control'
classpath = sourceSets.main.runtimeClasspath
main 'com.mypackage.SyncScript'
args Arrays.asList('jdbc:mysql://registry/db', 'com.mysql.jdbc.Driver', 'user', 'password').toArray()
}
// [3] Tell Gradle to invoke your Groovy script task.
defaultTasks 'runScript'Presto! It’s fairly trivial to write a Gradle build script that executes some arbitrary Groovy code. Since the preferred way to run Gradle these days is through a thin wrapper script, you can deliver this solution straight from your source control repo to anywhere without Gradle being installed.
In other words, the above is all you need to make Jenkins run a Groovy script whenever there’s a commit to its source control repository.
Groovy SQL
Now for the really neat part, the Groovy “sync” script itself. This script scans an arbitrary number of per-environment directories, scans an arbitrary number of per-application property files within each directory, and syncs those properties with a MySQL database table.
// Iterate through each per-environment directory
new File('config').eachDir { File environmentDirectory ->
// Iterate through each per-application properties file
environmentDirectory.eachFileMatch FileType.FILES, ~/.+\.properties/, { File applicationFile ->
def environment = environmentDirectory.name
def application = applicationFile.name.replace('.properties', '')
println "Processing properties for env: '$environment', application: '$application'"
// Parse the file into a java.util.Properties object
def properties = new Properties()
applicationFile.withInputStream { stream -> properties.load(stream) }
...
}
}Java 8 Streams have made this sort of thing more friendly and readable in pure-Java land, but it still can’t touch the simplicity of Groovy’s extensions to classes like File. The eachDir() and eachFileMatch() add-on methods make it easy to iterate through all of the directories, and scan for files having the extension “.properties“. The withInputStream() method helps us load the contents of each file into a java.util.Properties object with a one-liner.
Aside from extensions to java.io.File, Groovy offers its own groovy.sql.Sql class to facilitate JDBC operations. This reduces much of the boilerplate needed to construct a database query, and allows us to process its ResultSet within a closure:
database = groovy.sql.Sql.newInstance(jdbcUrl, jdbcUsername, jdbcPassword, jdbcDriver)
database.resultSetConcurrency = ResultSet.CONCUR_UPDATABLE
// Iterate through the properties, and sync MySQL
properties.entrySet().each {
def name = it.key
def value = it.value
def existingRecordQuery = '''
SELECT environment, service, property_name, property_value FROM environment_properties
WHERE environment = ? AND service = ? AND property_name = ?
'''
database.query(existingRecordQuery, [environment, service, name]) { ResultSet rs ->
if (rs.next()) {
def existingValue = rs.getString('property_value')
if (existingValue.equals(value)) {
// Existing property value is unchanged. No-op.
} else {
// Existing property value has changed. Update.
rs.updateString('property_value', value)
rs.updateRow()
}
} else {
// New property. Insert.
rs.moveToInsertRow()
rs.updateString('environment', environment)
rs.updateString('service', service)
rs.updateString('property_name', name)
rs.updateString('property_value', value)
rs.insertRow()
}
}
}
// TODO: Remove from the database properties that have
// been removed from the properties file.Several interesting things are happening here:
- On line 2, we change the concurrency setting to ResultSet.CONCUR_UPDATABLE. A lot of Java developers are unaware that Java even supports this!
- This setting allows you to update, insert, or delete rows in a ResultSet object, without having to construct additional JDBC statements. See examples of this happening on lines 20 and 29. Much of the convenience of an ORM, with the simplicity of raw JDBC!
- As you can see on lines 8-11, Groovy allows for multi-line string literals with triple quote marks. This makes it it much more readable to include long strings of SQL in your source code.
- On line 12, we see that groovy.sql.Sql allows you to execute a statement and deal with its result within a closure. One convenience is that the underlying JDBC statement will be closed automatically at the end.
Conclusion
This particular use case is very specific, but it showcases multiple concepts that are broadly useful in isolation. Groovy is a very powerful language, that may be welcome in environments where other alternatives are not. It’s native to Gradle, which is fast becoming the Java ecosystem’s most dominant build tool, and so Groovy is easy to leverage through your continuous integration server. Finally, Groovy comes with a full library of classes, and extensions to core Java classes, that really strip out much of the boilerplate and complexity of common tasks.
| Reference: | Easy Database Manipulation with Groovy and Gradle from our JCG partner Steve Perkins at the steveperkins.net blog. |
