S3 Gradle Plugin

Gradle plugin that uploads and downloads S3 objects.

Setup

New way:

plugins {
    id "com.github.mgk.gradle.s3" version "1.6.0"
}

Old way:

buildscript {
    repositories {
        maven {
            url "https://plugins.gradle.org/m2/"
        }
    }
    dependencies {
        classpath "gradle.plugin.com.github.mgk.gradle:s3:1.6.0"
    }
}

apply plugin: "com.github.mgk.gradle.s3"

Versioning

This project uses semantic versioning

See gradle plugin page for other versions.

Usage

Authentication

The S3 plugin searches for credentials in the same order as the AWS default credentials provider chain. Additionally you can specify a credentials profile to use by setting the project s3.profile property:

s3 {
    profile = 'my-profile'
}

Setting the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY is one way to provide your S3 credentials. See the AWS Docs for details on credentials.

Tasks

The following Gradle tasks are provided. See the tests for examples.

S3Upload

Upload a file to S3. Properties:

• bucket - S3 bucket to use (optional, defaults to the project s3 configured bucket)
• file - path of file to be uploaded
• key - key of S3 object to create.
• overwrite - (optional, default is false), if true the S3 object is created or overwritten if it already exists.

By default S3Upload does not overwrite the S3 object if it already exists. Set overwrite to true to upload the file even if it exists.

S3Download

Downloads one or more S3 objects. This task has two modes of operation: single file download and recursive download. Properties that apply to both modes:

• bucket - S3 bucket to use (optional, defaults to the project s3 configured bucket)

For a single file download:

• key - key of S3 object to download
• file - local path of file to save the download to

For a recursive download:

• keyPrefix - S3 prefix of objects to download
• destDir - local directory to download objects to

Nota Bene: recursive downloads create a sparse directory tree containing the full keyPrefix under destDir. So with an S3 bucket containing the object keys:

top/foo/bar
top/README

a recursive download:

task downloadRecursive(type: S3Download) {
  keyPrefix = "top/foo/"
  destDir = "local-dir"
}

results in this local tree:

local-dir/
└── foo
    └── bar

So only files under top/foo are downloaded, but their full S3 paths are appended to the destDir. This is different from the behavior of the aws cli aws s3 cp --recursive command which prunes the root of the downloaded objects. Use the flexible Gradle Copy task to prune the tree after downloading it. See example/build.gradle for an example.

Progress Reporting

Downloads report percentage progress at the gradle INFO level. Run gradle with the -i option to see download progress.

Development Notes

The example project pulls the build from Maven Local. Run ./gradlew publishToMavenLocal to deploy the plugin to the Maven Local repository. It also writes the current build to build/VERSION. The file example/build.gradle serves as tests and doc. The tests use the local build of the plugin.

The tests use a generated unique path in a test bucket to exercise all of the plugins features.

The test bucket has a AWS Object Expiration policy that removes objects older than one day automatically, so yay, no cleanup required.

The automated build uses an IAM access key that allows listBucket, getObject, and putObject on the test bucket only. The AWS credits configured as Travis environment variables. The access key id is not a secret so it is shown in the logs, while the secret access key is secret and not shown.

License