Gradle: process files inside a zip inside a tar - gradle

I have the following use case: I want to untar a tar file which contains a zip file. I then want to process files in the zip file (let's say I just want to filter out some file).
What I'm trying to do is something like:
task extractTar(type: Copy) {
from(tarTree("../a.tgz")) {
include '**/*.zip'
}
into "${buildDir}/tarOutput"
}
task unzipAndRezipZip(type: Zip, dependsOn: extractTar) {
archiveFileName = "rezipped.zip"
destinationDirectory
from(fileTree("${buildDir}/tarOutput").find { it.name.endsWith("zip") }) {
exclude '**/b.txt'
}
}
I've spent many many hours on this and every way I try to do it, it fails some way or another.
Expected Behavior
I can use the output files from a CopyTask in my ZipTask.
Current Behavior
path may not be null or empty string. path='null'
or
Circular dependency between the following tasks:
:unzipAndRezipZip
\--- :unzipAndRezipZip (*)
Steps to Reproduce
I have made a repo which has three examples of different things I tried: https://github.com/viktornordling/gradle-unzip-zip-from-tar-issue and clearly shows the repro steps for each example.
(I also put this in https://github.com/gradle/gradle/issues/16413, but I'm not sure how much love it will get there. Cross posting here to see if I can get some traction.)
Update: I would prefer not to use doLast in the solution as I feel that should not be needed. Usage of doLast is explicitly discouraged in the docs and not mentioned in the docs of CopyTask.

In example-4 from your repo https://github.com/viktornordling/gradle-unzip-zip-from-tar-issue you state that it almost works, but fails when you run clean unzipAndRezipZip in one go, with the error message Cannot expand ZIP...:
task unzipAndRezipZip(type: Zip) {
archiveFileName = "rezipped.zip"
destinationDir(file("${buildDir}/zipOutput"))
from(zipTree(tarTree("../a.tgz").find { it.name.endsWith("zip") })) {
exclude '**/b.txt'
}
}
You can fix that by changing the call to from() to a closure from { } to delay the execution of zipTree, i.e.:
task unzipAndRezipZip(type: Zip) {
archiveFileName = "rezipped.zip"
destinationDir(file("${buildDir}/zipOutput"))
from { zipTree(tarTree("../a.tgz").find { it.name.endsWith("zip") }) } {
exclude '**/b.txt'
}
}

Related

How do I get a task to execute of type Zip using 'dependsOn'?

I am trying to make my task zipGui execute on build target, but the only way I seem to be able to execute a task is if I remove the (type: Zip) from the task definition.
This snippet fails to execute the zipGui task at all:
task zipGui(type: Zip) {
doFirst {
println "==================== Zipping GUI components"
}
doLast {
FileTree zip = zipTree('assets/htdocs/gui.zip')
from 'assets/htdocs'
}
}
build.dependsOn zipGui
And this executes the zipGui but it doesn't know anything about how to zip files:
task zipGui {
doFirst {
println "==================== Zipping GUI components"
}
doLast {
FileTree zip = zipTree('assets/htdocs/gui.zip')
from 'assets/htdocs'
}
}
build.dependsOn zipGui
This is a stripped down fragment of the overall build.gradle.
How can I get zipGui to execute as a dependency of the build?
EDIT: here is more of the real build.gradle without me stripping things out to simplify the question:
task copyCert(type: Copy) {
from '../../../install'
into 'assets/certs/root'
include 'ca.pem'
doFirst {
println "==================== Copying root cert into assets"
}
}
task copyGui(dependsOn: copyCert, type: Copy) {
from '../../web/gui'
into 'assets/htdocs'
include '**/*.html'
include '**/*.css'
include '**/*.js'
include '**/*.wav'
include '**/*.tmpl'
include '**/*.png'
include '**/*.gif'
include '**/*.jpg'
exclude '**/*.DS_Store'
exclude '**/.gitignore'
exclude '**/.thumb'
exclude '**/build'
doFirst {
println "==================== Copying gui components into assets"
}
}
task zipGui(dependsOn: copyGui, type: Zip) {
FileTree zip = zipTree('assets/htdocs/gui.zip')
from 'assets/htdocs'
doFirst {
println "==================== Zipping GUI components"
}
}
Check details of zip task here: https://docs.gradle.org/current/dsl/org.gradle.api.tasks.bundling.Zip.html
task zipGui(type: Zip) {
archiveName = 'gui.zip'
from("$projectDir/../../../install") {
into 'assets/certs/root'
include 'ca.pem'
}
from("$projectDir/../../web/gui"){
into 'assets/htdocs'
exclude '**/*.DS_Store'
exclude '**/.gitignore'
exclude '**/.thumb'
exclude '**/build'
} // Just include or exclude is enough here as you indicated.
destinationDir(file("$buildDir/libs"))
}
Your task simply has nothing to do.
Gradle tasks are only executed, if they have something to do. If there is nothing to do (e.g. no files to zip), a task is skipped. There are multiple reasons for a task to have nothing to do. The main reason are up-to-date checks to prevent Gradle from doing the same thing what was done during the last invocation, at least as long as the task outputs are still available and the task inputs did not change.
However, in your specific case, the task has nothing to do, because at the time the task executes, no input files are specified at all. Files to include into the zip file can be added via from, which you do use. But, you use it inside a doLast closure, which is executed after the task actions (where the zipping takes place) were executed. Therefor, when running the zip process, the configuration did not take place.
You could use the regular configuration closure of the task or even the doFirst closure to configure your task:
task zipGui(type: Zip) {
from 'assets/htdocs'
}
// OR
task zipGui(type: Zip) {
doFirst {
from 'assets/htdocs'
}
}
Some more remarks:
The line FileTree zip = zipTree('assets/htdocs/gui.zip') does absolutely nothing, because the created file tree is not queried. You need to either access the files of the file tree manually or pass it to a task, e.g. your zip task: from zipTree('assets/htdocs/gui.zip')
In the second example, the task is executed because for normal tasks there are no automatic checks if there is any work to do, since a normal task can basically do everything the user makes it do.
I'm not quite sure, but I think the second example should throw an exception, because a normal task without a type does not provide a from method.
To get the status of each executed task and more insight generally, use the command line parameter --console=plain. The parameters -d and --stacktrace / --full-stacktrace may be used to get more information on errors.

Weird behaviour of a simple gradle copy task

I am completely baffled by the gradle behaviour of a Copy task into multiple directories.
I intend to copy all files from src/common into
target/dir1/ps_modules
target/dir2/ps_modules
target/dir3/ps_modules
Below is how my build.gradle looks:
project.ext.dirs = ["dir1", "dir2", "dir3"]
// ensures that ps_modules directory is created before copying
def ensurePsModulesDir() {
dirs.each {
def psModules = file("target/$it/ps_modules")
if (!(psModules.exists())) {
println("Creating ps_modules directory $psModules as it doesn't exist yet")
mkdir psModules
}
}
}
task copyCommons(type: Copy) {
doFirst {
ensurePsModulesDir()
}
from("src/common")
dirs.each {
into "target/$it/ps_modules"
}
}
The result of running the command ./gradlew copyCommons is completely weird.
The folder creation works as expected, however, the contents/files are copied only in the target/dir3/ps_modules directory. The rest two directories remain empty.
Any help would be appreciated.
Below is the screen grab of target directory tree once the job is run:
I think you want to do something like:
task copyCommons(type: Copy) {
dirs.each {
with copySpec {
from "src/common"
into "target/$it/ps_modules"
}
}
}
I think you can get rid of the ensurePsModulesDir() with this change
* edit *
it seems that the copy task is forcing us to set a destination dir. You might think that setting destinationDir = '.' is ok but it's used in up-to-date checking so likely the task will NEVER be considered up-to-date so will always run. I suggest you use project.copy(...) instead of a Copy task. Eg
task copyCommons {
// setup inputs and outputs manually
inputs.dir "src/common"
dirs.each {
outputs.dir "target/$it/ps_modules"
}
doLast {
dirs.each { dir ->
project.copy {
from "src/common"
into "target/$dir/ps_modules"
}
}
}
}
You can configure a single into for a task of type Copy. In this particular example gradle behaves as expected. Since dir3 is the last element on the list it is finally configured as a destination. Please have a look at this question - which you can find helpful. Also this thread might be helpful as well.

How can I define two different 'distribution' tasks in gradle?

The normal behavior of the distTar and distZip tasks from the application plugin in gradle seems to be to copy the contents of src/dist into the zip and tar files, but I have a subfolder in src/dist that I want to exclude from the default distribution, and include it for a new (extended) task, possibly to be called distZipWithJRE.
I have been able to exclude this folder in the default task as follows:
distributions.main {
contents {
from('build/config/main') {
into('config')
}
from('../src/dist') {
exclude('jre')
}
}
}
How can I define a second task that behaves just like the original (unmodified) task?
Using Gradle 4.8 I had to tweak the answer to use 'with' from CopySpec instead
distributions {
zipWithJRE {
baseName = 'zipWithJRE'
contents {
with distributions.main.contents
}
}
}
It seems that what you're looking for is in the docs. You need to leave current settings as is and for zipWithJRE create and configure custom distribution:
distributions {
zipWithJRE {
baseName = 'zipWithJRE'
contents {
from { distributions.main.contents }
}
}
}

how to download external files in gradle?

I have a gradle project which requires some data files available somewhere on the internet using http. The goal is that this immutable remote file is pulled once upon first build. Subsequent build should not download again.
How can I instruct gradle to fetch the given file to a local directory?
I've tried
task fetch(type:Copy) {
from 'http://<myurl>'
into 'data'
}
but it seems that copy task type cannot deal with http.
Bonus question: is there a way to resume a previously aborted/interrupted download just like wget -c does?
How about just:
def f = new File('the file path')
if (!f.exists()) {
new URL('the url').withInputStream{ i -> f.withOutputStream{ it << i }}
}
You could probably use the Ant task Get for this. I believe this Ant task does not support resuming a download.
In order to do so, you can create a custom task with name MyDownload. That can be any class name basically. This custom task defines inputs and outputs that determine whether the task need to be executed. For example if the file was already downloaded to the specified directory then the task is marked UP-TO-DATE. Internally, this custom task uses the Ant task Get via the built-in AntBuilder.
With this custom task in place, you can create a new enhanced task of type MyDownload (your custom task class). This task set the input and output properties. If you want this task to be executed, hook it up to the task you usually run via task dependencies (dependsOn method). The following code snippet should give you the idea:
task downloadSomething(type: MyDownload) {
sourceUrl = 'http://www.someurl.com/my.zip'
target = new File('data')
}
someOtherTask.dependsOn downloadSomething
class MyDownload extends DefaultTask {
#Input
String sourceUrl
#OutputFile
File target
#TaskAction
void download() {
ant.get(src: sourceUrl, dest: target)
}
}
Try like that:
plugins {
id "de.undercouch.download" version "1.2"
}
apply plugin: 'java'
apply plugin: 'de.undercouch.download'
import de.undercouch.gradle.tasks.download.Download
task downloadFile(type: Download) {
src 'http://localhost:8081/example/test-jar-test_1.jar'
dest 'localDir'
}
You can check more here: https://github.com/michel-kraemer/gradle-download-task
For me works fine..
The suggestion in Ben Manes's comment has the advantage that it can take advantage of maven coordinates and maven dependency resolution. For example, for downloading a Derby jar:
Define a new configuration:
configurations {
derby
}
In the dependencies section, add a line for the custom configuration
dependencies {
derby "org.apache.derby:derby:10.12.1.1"
}
Then you can add a task which will pull down the right files when needed (while taking advantage of the maven cache):
task deployDependencies() << {
String derbyDir = "${some.dir}/derby"
new File(derbyDir).mkdirs();
configurations.derby.resolve().each { file ->
//Copy the file to the desired location
copy {
from file
into derbyDir
// Strip off version numbers
rename '(.+)-[\\.0-9]+\\.(.+)', '$1.$2'
}
}
}
(I learned this from https://jiraaya.wordpress.com/2014/06/05/download-non-jar-dependency-in-gradle/).
Using following plugin:
plugins {
id "de.undercouch.download" version "3.4.3"
}
For a task which has the purpose of only downloading
task downloadFile(type: Download) {
src DownloadURL
dest destDir
}
For including download option into your task:
download {
src DownloadURL
dest destDir
}
For including download option with multiple downloads into your task:
task downloadFromURLs(){
download {
src ([
DownloadURL1,
DownloadURL2,
DownloadURL3
])
dest destDir
}
}
Hope it helped :)
just now ran into post on upcoming download task on gradle forum.
Looks like the perfect solution to me.. Not (yet) available in an official gradle release though
Kotlin Version of #Benjamin's Answer
buildscript {
repositories {
jcenter()
google()
}
dependencies {
classpath("com.android.tools.build:gradle:4.0.1")
}
}
tasks.register("downloadPdf"){
val path = "myfile.pdf"
val sourceUrl = "https://file-examples-com.github.io/uploads/2017/10/file-sample_150kB.pdf"
download(sourceUrl,path)
}
fun download(url : String, path : String){
val destFile = File(path)
ant.invokeMethod("get", mapOf("src" to url, "dest" to destFile))
}

Extract specific JARs from dependencies

I am new to gradle but learning quickly. I need to get some specific JARs from logback into a new directory in my release task. The dependencies are resolving OK, but I can't figure out how, in the release task, to extract just logback-core-1.0.6.jar and logback-access-1.0.6.jar into a directory called 'lib/ext'. Here are the relevant excerpts from my build.gradle.
dependencies {
...
compile 'org.slf4j:slf4j-api:1.6.4'
compile 'ch.qos.logback:logback-core:1.0.6'
compile 'ch.qos.logback:logback-classic:1.0.6'
runtime 'ch.qos.logback:logback-access:1.0.6'
...
}
...
task release(type: Tar, dependsOn: war) {
extension = "tar.gz"
classifier = project.classifier
compression = Compression.GZIP
into('lib') {
from configurations.release.files
from configurations.providedCompile.files
}
into('lib/ext') {
// TODO: Right here I want to extract just logback-core-1.0.6.jar and logback-access-1.0.6.jar
}
...
}
How do I iterated over the dependencies to locate those specific files and drop them in the lib/ext directory created by into('lib/ext')?
Configurations are just (lazy) collections. You can iterate over them, filter them, etc. Note that you typically only want to do this in the execution phase of the build, not in the configuration phase. The code below achieves this by using the lazy FileCollection.filter() method. Another approach would have been to pass a closure to the Tar.from() method.
task release(type: Tar, dependsOn: war) {
...
into('lib/ext') {
from findJar('logback-core')
from findJar('logback-access')
}
}
def findJar(prefix) {
configurations.runtime.filter { it.name.startsWith(prefix) }
}
It is worth nothing that the accepted answer filters the Configuration as a FileCollection so within the collection you can only access the attributes of a file. If you want to filter on the dependency itself (on group, name, or version) rather than its filename in the cache then you can use something like:
task copyToLib(type: Copy) {
from findJarsByGroup(configurations.compile, 'org.apache.avro')
into "$buildSrc/lib"
}
def findJarsByGroup(Configuration config, groupName) {
configurations.compile.files { it.group.equals(groupName) }
}
files takes a dependencySpecClosure which is just a filter function on a Dependency, see: https://gradle.org/docs/current/javadoc/org/gradle/api/artifacts/Dependency.html

Resources