The question is specific, but it's more of a general 'how to do this in gradle' question.
I have a demo java web app that I can run using the gretty plugin. I would like to selectively control whether a javaagent is applied to the jvmArgs of the gretty process based on a command line flag. The agent jar location is known by getting its path from a dummy configuration:
configurations {
agent
}
dependencies {
...
agent group: 'com.foo', name: 'foo-agent', version: '1.0'
}
I know I can access the jar file location using something like:
project.configurations.agent.find { it.name.startsWith("foo-agent") }
How can I selectively apply that to the gretty jvmArgs configuration based on a command line property such as
gradle -PenableAgent
I ended up solving this by creating a task and simply calling it before I run the war:
task agent {
doFirst {
def agentJar = project.configurations.agent.find { it.name.startsWith("foo-agent") }
gretty.jvmArgs << "-javaagent:" + agentJar
}
}
Then I can simply call:
gradle agent appRunWar
In my project I use Spring Instrument as java agent so this was my solution.
You can make appRun task dependent on agent task then no additional gradle run parameter needed.
dependencies {
...
agent 'org.springframework:spring-instrument:4.2.4.RELEASE'
}
configurations {
dev
agent
}
gretty {
...
contextPath = '/'
jvmArgs=[]
springBoot = true
...
}
task agent {
doFirst {
def agentJar = project.configurations.agent.find{it.name.contains("spring-instrument") }
gretty.jvmArgs << "-javaagent:" + agentJar
}
}
project.afterEvaluate {
tasks.appRun.dependsOn agent
}
Related
I have a task with name test and code is as below :
tasks {
"test"(Test::class) {
useJUnitPlatform {
excludeTags = setOf("e2e", "integration")
}
}
When execute this task with gradle command ./gradlew test --info, gradle is scanning all the modules of my project and generating a configuration with has some tasks from module named data-export-ui-kjs.
I want gradle to exclude the tasks from data-export-ui-kjs module while configuring & executing test task. I have used below code to achieve this but its not successful
gradle.taskGraph.whenReady {
TaskExecutionGraphListener { graph ->
if(it.name.contains("data-export-ui-kjs")) {
it.enabled = false
}
}
}
Kindly help me to get this done. Thanks in advance
Working as expected. Modified the code. Its syntax issue itseems
When performing gradle clean and then gradle swagger a ClassNotFoundException is thrown. If gradle swagger is then run again (basically after the api build is done in previous run), it works fine.
build.gradle looks as below:
buildscript {
repositories {
maven { url hydraMavenRepo }
maven { url hydraPluginsRepo }
}
dependencies {
classpath "com.github.kongchen:swagger-maven-plugin:3.1.4"
}
}
apply plugin: 'java'
configurations {
addclasspath
}
dependencies {
addclasspath files(project(':api:desktop-api').configurations['runtime'].files)
addclasspath files(project(':api:desktop-api').sourceSets['main'].output)
addclasspath files(project(':api:desktop-api').sourceSets.main.output.classesDir)
runtime project(':api:desktop-api')
}
sourceSets {
main {
runtimeClasspath += files(project(':api:desktop-api').sourceSets['main'].output)
runtimeClasspath += files(project(':api:desktop-api').sourceSets['main'].output.classesDir)
runtimeClasspath += files(project(':api:desktop-api').configurations['runtime'].files)
}
}
import com.github.kongchen.swagger.docgen.mavenplugin.ApiDocumentMojo
import com.github.kongchen.swagger.docgen.mavenplugin.ApiSource
import io.swagger.models.Info
task swagger(dependsOn: [':api:desktop-api:build']) {
doLast {
logger.info 'Swagger GenDoc...'
project.file(reportsDir).mkdirs()
// a trick to have all needed classes in the classpath
def customClassLoader = new GroovyClassLoader()
buildscript.configurations.classpath.each {
//println it.toURI().toURL()
customClassLoader.addURL(it.toURI().toURL())
}
configurations.addclasspath.each {
customClassLoader.addURL(it.toURI().toURL())
}
// the same settings as in the swagger-maven-example/pom.xml
final ApiDocumentMojo mavenTask = Class.forName('com.github.kongchen.swagger.docgen.mavenplugin.ApiDocumentMojo', true, customClassLoader).newInstance(
apiSources: [
new ApiSource(
springmvc: false,
locations: ['com/vmware/vdi/hydra'],
schemes: ['http', 'https'],
host: 'vmware.com',
basePath: '/api',
info: new Info(
title: "Hydra DS-REST API's",
version: 'v100',
description: "Hydra DS-REST API's",
),
swaggerDirectory: reportsDir
)
]
)
mavenTask.execute()
logger.info 'Swagger GenDoc task is completed'
}
}
You have several flaws in your build script.
You should not depend on built things in build script dependencies. This is a hen and egg problem. You need to execute the build to get the classes that are necessary to execute the build. This cannot work reliably, if at all.
Instead you should declare them as dependencies outside the buildscript block. The buildscript block is only for dependencies that are needed by your build script to run itself, like Gradle Tasks and Gradle Plugins or classes you use during the build, like the swagger-maven-plugin stuff which is correct in the buildscript block.
Besides that, you execute part of your swagger stuff (the instanciation, execution and printing) during the configuration phase instead of during the execution phase. Everything you do in a task closure, but outside any doFirst or doLast blocks is run during the configuration phase, when the build is configured and thus always, no matter what tasks you actually want to execute and no matter whether the task may already be up-to-date or not. For the up-to-date check to work and save your time you need to declare all inputs like files and properties that might change between executions and all outputs that you generate, then Gradle can do its magic to only execute the task when actually necessary.
Also you should not use println in build scripts. That is like using System.out.println in Java programs. Instead you should use the provided logging facility directly, e. g. doing logger.info 'Swagger GenDoc task is completed'.
buildscript.classloader is what I was looking for.
Below is the code that works:
buildscript {
repositories {
maven { url mavenRepo }
}
dependencies {
classpath "com.github.kongchen:swagger-maven-plugin:3.1.4"
}
}
import com.github.kongchen.swagger.docgen.mavenplugin.ApiDocumentMojo
import com.github.kongchen.swagger.docgen.mavenplugin.ApiSource
import io.swagger.models.Info
task swagger(dependsOn: ':api:build') {
doLast {
logger.info 'Swagger GenDoc...'
project.file(<dir>).mkdirs()
FileCollection apiRuntimeFiles = files(project(':api').configurations['runtime'].files)
apiRuntimeFiles.each {
buildscript.classLoader.addURL(it.toURI().toURL())
}
FileCollection apiClassFiles =files(project(':api').sourceSets['main'].output)
apiClassFiles.each {
buildscript.classLoader.addURL(it.toURI().toURL())
}
final ApiDocumentMojo mavenTask = Class.forName('com.github.kongchen.swagger.docgen.mavenplugin.ApiDocumentMojo', true, buildscript.classLoader).newInstance(
apiSources: [
new ApiSource(
springmvc: false,
locations: ['<loc>'],
schemes: ['http', 'https'],
host: '<host>',
basePath: '/api',
info: new Info(
title: "REST API's",
version: 'v1',
description: "REST API's",
),
swaggerDirectory: <dir>
)
]
)
mavenTask.execute()
logger.info 'Swagger GenDoc task is completed'
}
}
I am in love with JBoss TattleTale. Typically, in my Ant builds, I follow the docs to define the Tattletale tasks and then run them like so:
<taskdef name="report"
classname="org.jboss.tattletale.ant.ReportTask"
classpathref="tattletale.lib.path.id"/>
...
<tattletale:report source="${src.dir]" destination="${dest.dir}"/>
I am now converting my builds over to Gradle and am struggling to figure out how to get Tattletale running in Gradle. There doesn't appear to be a Gradle-Tattletale plugin, and I'm not experienced enough with Gradle to contribute one. But I also know that Gradle can run any Ant plugin and can also executing stuff from the system shell; I'm just not sure how to do this in Gradle because there aren't any docs on this (yet).
So I ask: How do I run the Tattletale ReportTask from inside a Gradle build?
Update
Here is what the Gradle/Ant docs show as an example:
task loadfile << {
def files = file('../antLoadfileResources').listFiles().sort()
files.each { File file ->
if (file.isFile()) {
ant.loadfile(srcFile: file, property: file.name)
println " *** $file.name ***"
println "${ant.properties[file.name]}"
}
}
}
However, no where in here do I see how/where to customize this for Tattletale and its ReportTask.
The following is adapted from https://github.com/roguePanda/tycho-gen/blob/master/build.gradle
It bypasses ant and directly invokes the Tattletale Java class.
It was changed to process a WAR, and mandates a newer javassist in order to handle Java 8 features such as lambdas.
configurations {
tattletale
}
configurations.tattletale {
resolutionStrategy {
force 'org.javassist:javassist:3.20.0-GA'
}
}
dependencies {
// other dependencies here...
tattletale "org.jboss.tattletale:tattletale:1.2.0.Beta2"
}
task createTattletaleProperties {
ext.props = [reports:"*", enableDot:"true"]
ext.destFile = new File(buildDir, "tattletale.properties")
inputs.properties props
outputs.file destFile
doLast {
def properties = new Properties()
properties.putAll(props)
destFile.withOutputStream { os ->
properties.store(os, null)
}
}
}
task tattletale(type: JavaExec, dependsOn: [createTattletaleProperties, war]) {
ext.outputDir = new File(buildDir, "reports/tattletale")
outputs.dir outputDir
inputs.files configurations.runtime.files
inputs.file war.archivePath
doFirst {
outputDir.mkdirs()
}
main = "org.jboss.tattletale.Main"
classpath = configurations.tattletale
systemProperties "jboss-tattletale.properties": createTattletaleProperties.destFile
args([configurations.runtime.files, war.archivePath].flatten().join("#"))
args outputDir
}
The previous answers either are incomplete or excessively complicated. What I did was use the ant task from gradle which works fine. Let's assume your tattletale jars are beneath rootDir/tools/...
ant.taskdef(name: "tattleTaleTask", classname: "org.jboss.tattletale.ant.ReportTask", classpath: "${rootDir}/tools/tattletale-1.1.2.Final/tattletale-ant.jar:${rootDir}/tools/tattletale-1.1.2.Final/tattletale.jar:${rootDir}/tools/tattletale-1.1.2.Final/javassist.jar")
sources = "./src:./src2:./etcetera"
ant.tattleTaleTask(
source: sources,
destination: "tattleTaleReport",
classloader: "org.jboss.tattletale.reporting.classloader.NoopClassLoaderStructure",
profiles: "java5, java6",
reports: "*",
excludes: "notthisjar.jar,notthisjareither.jar,etcetera.jar"
){
}
So the above code will generate the report beneath ./tattleTaleReport. It's that simple. The annoyance is that the source variable only accepts directories so if there are jars present in those directories you do not wish to scan you need to add them to the excludes parameter.
I have a a version property in my gradle.properties file that gives the version I am building. I have a task in the build called release that if present in the task graph will upload to the snapshot repo. However what is happening is that even though I include the release task in the build tasks, snapshot is not appended to my version property when uploadArchives runs so it attempts to upload to the wrong repository and fails. The when ready runs, but it does not seem to run before uploadArchives. Can anyone explain what is happening here?
uploadArchives {
repositories {
ivy {
credentials {
username nexusUser
password nexusPassword
}
if (version.endsWith("-SNAPSHOT")) {
url nexusSnapshotRepository
} else {
url nexusReleaseRepository
}
}
}
}
gradle.taskGraph.whenReady {taskGraph ->
if (!taskGraph.hasTask(release)) {
version = version + '-SNAPSHOT'
}
println "release task not included - version set to $version"
}
task release(){
doLast{
println "Releasing"
}
}
This is very similar to the example on the gradle site so I don't see what is going wrong.
http://www.gradle.org/docs/current/userguide/tutorial_using_tasks.html
The script is checking the project.version value in the configuration phase (not when the task executes), but only modifying it after the task execution graph has been built. One way to fix this is to override the repository url from inside the taskGraph.whenReady callback:
uploadArchives {
repositories {
ivy {
name "nexus"
url nexusReleaseRepository
...
}
}
}
gradle.taskGraph.whenReady { taskGraph ->
if (!taskGraph.hasTask(release)) {
version += '-SNAPSHOT'
uploadArchives.repositories.nexus.url nexusSnapshotRepository
// ps: println has to go inside here
}
}
I'm trying to create a test task rule using the example provided in the grails gradle doc but I keep getting "a task with that name already exists" error.
My build script is as follows:
import org.grails.gradle.plugin.tasks.* //Added import here else fails with "Could not find property GrailsTestTask"
buildscript {
repositories {
jcenter()
}
dependencies {
classpath "org.grails:grails-gradle-plugin:2.0.0"
}
}
version "0.1"
group "example"
apply plugin: "grails"
repositories {
grails.central() //creates a maven repo for the Grails Central repository (Core libraries and plugins)
}
grails {
grailsVersion = '2.3.5'
groovyVersion = '2.1.9'
springLoadedVersion '1.1.3'
}
dependencies {
bootstrap "org.grails.plugins:tomcat:7.0.50" // No container is deployed by default, so add this
compile 'org.grails.plugins:resources:1.2' // Just an example of adding a Grails plugin
}
project.tasks.addRule('Pattern: grails-test-app-<phase>') { String taskName ->
println tasks //shows grails-test-app-xxxxx task. Why?
//if (taskName.startsWith('grails-test-app') && taskName != 'grails-test-app') {
// task(taskName, type: GrailsTestTask) {
// String testPhase = (taskName - 'grails-test-app').toLowerCase()
// phases = [testPhase]
// }
//}
}
Running $gradle grails-test-integration
or in fact anything of the form $gradle grails-test-app-xxxxxxxx yields the error "Cannot add task 'gradle grails-test-app-xxxxxxxx as a task with that name already exists".
Can someone please advise how I can resolve this error? Thanks.
If you don't mind overriding the task created by the plugin, you might want to try
task(taskName, type: GrailsTestTask, overwrite: true)
In general, when using task rules that can be called multiple times (for instance if you have multiple tasks depending on a task eventually added by your rules), I use the following test before actually creating the task:
if (tasks.findByPath(taskName) == null) {tasks.create(taskName)}
This will call the task() constructor only if this task name does not exists.