Skip to content

Latest commit

 

History

History
652 lines (528 loc) · 28.2 KB

common-plugin.md

File metadata and controls

652 lines (528 loc) · 28.2 KB
Raw

Cognifide logo

Gradle AEM Plugin

Common plugin

About

plugins {
    id("com.cognifide.aem.common")
}

Applied transparently by other plugins. Provides AEM extension to build script / Gradle AEM DSL which consists of instance definitions, common configuration, methods for controlling local instances.

It does not provide any tasks but is a base for implementing them on our own.
Apply other plugins to provide pre-defined, ready-to-use tasks.

Instance conventions

  • Instance name is a combination of ${environment}-${purposeId} e.g local-author, integration-publish etc.
  • Instance purposeId is an instance purpose identifier and must start with the prefix author or publish. Sample valid names: author, author1, author2, author-master and publish, publish1 publish2, etc.
  • Instance location indicates where instance is physically located, possible values are: local and remote. Local means that instance could be created by the plugin automatically under the local file system.
  • Only instances with location local are considered in command instanceSetup, instanceCreate, instanceUp etc (that comes from the plugin com.cognifide.aem.instance.local).
  • All instances with any location are considered in commands CRX package deployment-related like instanceProvision, packageDeploy, packageUpload, packageInstall etc.

Instances could be defined in two ways, via:

  • file gradle.properties - recommended approach, by properties convention.
  • build script - dynamic & more customizable approach.

Defining instances by a properties file

The configuration could be specified through gradle.properties file using dedicated syntax.

instance.$ENVIRONMENT-$ID.$PROP_NAME=$PROP_VALUE

Part Possible values Description
$ENVIRONMENT local, int, stg etc Environment name.
$ID author, publish, publish2, etc Combination of AEM instance type and semantic suffix useful when more than one of instance of same type is being configured.
$PROP_NAME=$PROP_VALUE Local instances: enabled, httpUrl=http://admin:admin@localhost:4502
type=local
password=foo
runModes=nosamplecontent
jvmOpts=-server -Xmx2048m -XX:MaxPermSize=512M -Djava.awt.headless=true, startOpts=...
debugPort=24502.

Remote instances: enabled, httpUrl, type=remote, user, password.		 Run modes, JVM opts and start opts should be comma delimited.

Default remote instances defined via properties (below lines are optional):

instance.local-author.httpUrl=http://localhost:4502
instance.local-publish.httpUrl=http://localhost:4503

Example of defining multiple remote instances (that could be filtered):

instance.int-author.httpUrl=http://author.aem-integration.company.com
instance.int-publish.httpUrl=http://aem-integration.company.com
instance.stg-author.httpUrl=http://author.aem-staging.company.com
instance.stg-publish.httpUrl=http://aem-staging.company.com

Example for defining remote instance with credentials separated:

instance.test-author.httpUrl=http://author.aem-integration.company.com
instance.test-author.user=foo
instance.test-author.password=bar

Example for defining remote instance with credentials details included in URL:

instance.test-author.httpUrl=http://foo:[email protected]

Example for defining local instances (created on the local file system):

instance.local-author.httpUrl=http://localhost:4502
instance.local-author.type=local
instance.local-author.debugPort=14502
# debugAddress can be used to enable access to debug port from outside the local network (Java 9+)
instance.local-author.debugAddress=*
instance.local-author.runModes=nosamplecontent
instance.local-author.jvmOpts=-server -Xmx1024m -XX:MaxPermSize=256M -Djava.awt.headless=true

instance.local-publish.httpUrl=http://localhost:4503
instance.local-publish.type=local
instance.local-publish.debugPort=14503
instance.local-publish.debugAddress=*
instance.local-publish.runModes=nosamplecontent
instance.local-publish.jvmOpts=-server -Xmx1024m -XX:MaxPermSize=256M -Djava.awt.headless=true

Notice! Remember to define also AEM source files.

Defining instances by a build script

Example usage below. The commented value is an effective instance name.

aem {
    instance {
        local("http://localhost:4502") // local-author
        local("http://localhost:4502") { // local-author
            password = "admin"
            id = "author"
            debugPort = 14502 
        }

        local("http://localhost:4503") // local-publish
        local("http://localhost:4503") { // local-publish
            password = "admin"
            id = "publish"
            debugPort = 14503
        } 

        remote("http://192.168.10.1:4502") { // integration-author1
            user = "user1" 
            password = "password2"
            env = "integration"
            id = "author1"
        } 
        remote("http://192.168.10.1:8080") { // integration-author2
            user = "user1" 
            password = "password2"
            env = "integration"
            id = "author2"
        } 
        remote("http://192.168.10.2:4503") { // integration-publish1
            user = "user2"
            password = "password2"
            env = "integration"
            id = "publish1"
        } 
        remote("http://192.168.10.2:8080") { // integration-publish2
            user = "user2"
            password = "password2"
            env = "integration"
            id = "publish2"
        } 
    }
}

Instance filtering

When there are defined named AEM instances: local-author, local-publish, integration-author and integration-publish,
then it is possible to:

  • deploy (or satisfy) CRX package(s),
  • tail logs,
  • checkout JCR content,

with taking into account:

  • type of environment (local, integration, staging, etc),
  • type of AEM instance (author / publish),

by filtering instances by names, e.g:

gradlew packageDeploy -Pinstance.name=integration-*
gradlew packageDeploy -Pinstance.name=*-author
gradlew packageDeploy -Pinstance.name=local-author,integration-author

The default value of that instance name filter is ${env}-*, so that typically local-*.
Environment value comes from the system environment variable ENV or property env.

To deploy only to the author or publish instances:

gradlew packageDeploy -Pinstance.author
gradlew packageDeploy -Pinstance.publish

To deploy only to instances specified explicitly:

gradlew packageDeploy -Pinstance.list=[http://admin:admin@localhost:4502,http://admin:admin@localhost:4503]

Instance URLs must be delimited by a colon.
Remember to encode instance user & password properly.

Disabling defined instances

Sometimes, it might be useful to work only with AEM author instance (skipping publish) to optimize CPU usage / consume less workstation resources. Another case could be to enable some extra AEM publish instance (like so called preview publish) only temporarily.

Controlling defined instances availability could be done by additional property instance.$NAME.enabled for instances defined by properties.

instance.local-author.httpUrl=http://localhost:4502
instance.local-author.type=local
instance.local-author.enabled=false

Still, such disabled AEM instances could be useful in scripting using Gradle AEM DSL, so they are accessible via method aem.instance(name), but not via aem.namedInstance(name), aem.instances, aem.authorInstance, aem.publishInstance which are relying on instance filtering.

Instance URL credentials encoding

Remember to encode instance credentials (user & password) when passing it via properties: instance.$INSTANCE_NAME.httpUrl or instance.list=[$INSTANCE_HTTP_URL1,$INSTANCE_HTTP_URL2,...].

For example, let's assume that instance is created by instance plugin using the following properties:

instance.local-author.httpUrl=http://localhost:4502
instance.local-author.type=local
instance.local-author.password=gxJMge@6F5ZV#s9j

The password is generated by e.g strong password generator so it has special URL characters.
Then to be able to use this created instance but remotely and via instance.list property on CI, then it should look as below:

sh gradlew :aem:assembly:full:packageDeploy -Pinstance.list=[http://admin:gxJMge%406F5ZV%[email protected]:4502]

Notice that when a password is specified as separate property instance.$INSTANCE_NAME.password, then it does not need to be encoded.
Otherwise, password value must be encoded by e.g online URL encoder.

Implementing tasks

Most of the built-in tasks logic is based on object aem of type AemExtension.
It provides concise AEM related API for accessing AEM configuration, synchronizing with AEM instances via specialized instance services of aem.sync
to make tasks implementation a breeze. The options for automating things around AEM are almost unlimited.

Instance services

While implementing custom AEM tasks, mix usages of following instance services:

  • http - InstanceHttpClient - Provides extremely easy to use HTTP client designed especially to be used with AEM (covers basic authentication, allows to use only relative paths instead of full URLs, etc)
  • packageManager - PackageManager - Allows communicating with the CRX Package Manager.
  • osgiFramework - OsgiFramework - Controls OSGi framework using Apache Felix Web Console endpoints.
  • repository - Repository - Allows communicating with JCR Content Repository.
  • workflowManager - WorkflowManager - Allows to temporarily toggle (enable or disabled) change workflow launcher state e.g to disable DAM assets regeneration while deploying the CRX package.
  • groovyConsole - GroovyConsole - Allows to execute Groovy code/scripts on AEM instance having Groovy Console CRX package installed.
  • status - Status - Allows to read statuses available at Apache Felix Web Console.
  • crx - Crx - Allows reading available node types of JCR repository / AEM instance.

Defining CRX package via code then downloading and sharing it using an external HTTP endpoint

The below snippet could be used to automatize the creation of production content backups.

aem {
    tasks {
        register("backupProductionAuthor") {
            doLast {
                val pkg = namedInstance("prod-author").sync {
                    downloadPackage {
                        group = "example"
                        name = "backup"
                        description = "Backup of content, tags and DAM"
                        archiveName = "backup-author.zip"
                        filters(
                                "/content/cq:tags/example",
                                "/content/example",
                                "/content/dam/example"
                        )
                    }
                }

                http {
                    basicUser = "foo"
                    basicPassword = "bar"
                    postMultipart("http://my-aem-backup-service.com/package/upload", mapOf("file" to pkg)) 
                }
            }
        }
    }
}

Calling AEM endpoints / making any HTTP requests

To make an HTTP request to some AEM endpoint (servlet) simply write:

aem {
    tasks {
        register("runHealthCheck") {
            doLast {
                syncInstances {
                    http {
                        get("/bin/example/healthCheck") { checkStatus(it, 200) }
                    }
                }
            }
        }
    }
}

There are unspecified AEM instances as the parameter for the method syncInstances, so that instances matching default filtering will be used.

The fragment { checkStatus(it, 200) } could be even omitted because, by default, sync API checks the status code that it belongs to the range [200,300).

To parse endpoint response as JSON (using JsonPath), simply write:

aem {
    tasks {
        register("runHealthCheck") {
            doLast {
                syncInstances {
                    http {
                        val json = get("/bin/example/healthCheck") { asJson(it) }
                        val status = json.read("status") as String

                        if (status != "OK") {
                            throw GradleException("Health check failed on: $instance because status '$status' detected.")
                        }
                    }
                }
            }
        }
    }
}

There are also available convenient methods asStream, asString to be able to process endpoint responses.

Downloading the CRX package from external HTTP endpoint and deploying it on desired AEM instances

Below snippet could be used to automatize recovery from content backups (e.g for production or to replicate production content to test environment).

aem {
    tasks {
        register("deployProductionContent") {
            doLast {
                val instances = listOf(
                        instance("http://user:[email protected]") // URL specified directly, could be parametrized by some gradle command line property
                        // namedInstance("local-publish") // reused AEM instance defined in 'gradle.properties'
                )
                val pkg = httpFile { download("https://company.com/aem/backups/example-1.0.0-201901300932.backup.zip") }

                sync(instances) { 
                    packageManager.deploy(pkg) 
                }
            }
        }
    }
}

Transferring live content CRX package to local AEM instances

Task contentSync only downloads and deploys packages when a new live content package will become available and it is not yet deployed on local AEM instances.
The example below assumes a dynamic instance definition based on Gradle properties.

tasks {
    register("contentSync") {
        dependsOn(":requireProps")
        doLast {
            val sourceInstance = aem.instance(findProperty("contentSyncInstanceUrl")?.toString().orEmpty()).apply {
                user = findProperty("contentSyncUser")?.toString().orEmpty()
                password = findProperty("contentSyncPassword")?.toString().orEmpty()
                sync { packageManager.requireAvailable() }
            }
            val sourcePkg = sourceInstance.sync {
                packageManager.all.asSequence()
                        .filter { it.group == "content-sync" }
                        .maxBy { it.lastTouched }
                        ?: throw GradleException("Cannot find content sync package on $instance!")
            }
            val upToDate = aem.instances.all { it.sync { packageManager.contains(sourcePkg) } }
            if (upToDate) {
                logger.lifecycle("Most recent 'content-sync' package is available on all instances.")
            } else {
                val targetDir = file("build/contentSync").apply {
                    deleteRecursively()
                    mkdirs()
                }
                val targetFile = sourceInstance.sync { packageManager.downloadTo(sourcePkg, targetDir) }
                aem.sync { packageManager.deploy(targetFile) }

                logger.lifecycle("Most recent 'content-sync' package has been deployed to all instances.")
            }
        }
    }
}

Working with the content repository (JCR)

To make changes in the AEM content repository, use the Repository instance service which is a part of instance sync tool.

For example, to migrate pages even without using Groovy Console deployed on some AEM instance, simply write:

aem {
    tasks {
        register("migratePages") {
            description = "Migrates pages to new component"
            doLast {
                syncInstances {
                    repository {
                        node("/content/example")
                            .traverse()
                            .filter { it.type == "cq:PageContent" && it.properties["sling:resourceType"] == "example/components/basicPage" }
                            .forEach { page ->
                                logger.info("Migrating page: ${page.path}")
                                page.saveProperty("sling:resourceType", "example/components/advancedPage")
                            }
                    }
                }
            }
        }
    }
}

To create new or update existing nodes, write:

aem {
    tasks {
        register("setupReplicationAgents") {
            description = "Corrects publish replication agent transport URI"
            doLast {
                syncInstances {
                    repository {
                        node("/etc/some-tool/config", mapOf( // shorthand for 'node(path).save(props)'
                            "key" to "value",
                            "flag" to false,
                        ))
                    }
                }
            }
        }
    }
}

Under the hood, the repository service is using only AEM built-in Sling Post Servlet so it should work with all AEM instances that are having that servlet accessible (which reflects default AEM configuration).

Executing code on AEM runtime

It is also possible to easily execute any code on AEM runtime using the Groovy Console.
Assuming that on AEM instances there is already installed Groovy Console e.g via task instanceProvision, then it is possible to use the GroovyConsole instance service.

aem {
    provisioner {
        deployPackage("https://github.com/icfnext/aem-groovy-console/releases/download/12.0.0/aem-groovy-console-12.0.0.zip")
    }
    tasks {
        register("generatePosts") {
            doLast {
                syncInstances {
                    groovyConsole.evalCode("""
                        def postsService = getService("com.company.example.aem.sites.services.posts.PostsService")

                        println postsService.randomPosts(5)
                    """)
                    // groovyConsole.evalScript("posts.groovy") // if script above moved to 'aem/gradle/groovyScript/posts.groovy'
                }
            }
        }
    }
}

Controlling OSGi bundles, components, and configurations

Simply use the instance service: OsgiFramework.

To restart some bundle after deploying a CRX package, write:

aem {
    tasks {
        packageDeploy {
            doLast {
                syncInstances {
                    osgiFramework.restartBundle("com.adobe.cq.dam.cq-scene7-imaging")
                }           
            }       
        }
    }
}

To disable specific OSGi component by its PID value and only on publish instances, write:

aem {
    tasks {
        register("instanceSecure") {
            doLast {
                sync(publishInstances) {
                    osgiFramework.disableComponent("org.apache.sling.jcr.davex.impl.servlets.SlingDavExServlet")
                    // osgiFramework.stopBundle("org.apache.sling.jcr.webdav")
                }
            }
        }
    }
}

To configure specific OSGi service by its PID:

aem {
    tasks {
        register("enableCrx") {
            doLast {
                sync(publishInstances) {
                    osgiFramework.configure("org.apache.sling.jcr.davex.impl.servlets.SlingDavExServlet", mapOf(
                        "alias" to "/crx/server",
                        "dav.create-absolute-uri" to true,
                        "dav.protectedhandlers" to "org.apache.jackrabbit.server.remoting.davex.AclRemoveHandler"
                    ))
                }
            }
        }
    }
}

All CRUD methods for manipulating OSGi configurations are available. Also for configuration factories.

Controlling workflows

Simply use Workflow Manager instance service.

Workflows can be either enabled or disabled by:

aem {
    tasks {
        register("setupWorkflows") {
            doLast {
                syncInstances {
                    workflowManager.workflow("update_asset_create").disable()
                    workflowManager.workflow("update_asset_mod").disable()

                    workflowManager.workflows("dam_asset").forEach { it.enable() } // reverts above using shorthand alias
                }
            }        
        }   
    }   
}

Also, it is possible to enable or disable workflows only for a particular action to be performed:

    workflowManager.toggleTemporarily("dam_asset", false) {
        packageManager.deploy(file("my-package.zip"))
    }

It is also possible to mix enabling and disabling workflows:

    workflowManager.toggleTemporarily(mapOf(
        "update_asset_create" to false,
        "update_asset_mod" to false
        "update_asset_custom" to true 
    )) {
        // ...
    }

Running Docker image based tools

Regardless type of Docker runtime used (Desktop or Toolbox), running any Docker images is super-easy via Gradle AEM DSL methods aem.dockerRun() or aem.dockerDaemon().
Under the hood, GAP ensures that volume paths passed to Docker are correct. Also correctly handles streaming process output,
allows to set expected exit codes and other useful options.

Consider following task registration:

tasks {
    register("runTool")
        doLast {
            aem.runDocker {
                operation("Runnning Docker based tool'")
                image = "any-vendor/any-image"
                volume(file("resources"), "/resources")
                port(8080, 80)
                command = "<any command>"
            }
        }
    }
}

Then running any Docker image based tool could be simply achieved via running CLI command:

sh gradlew runTool.

What is more, it is possible to run any Docker based tool as a daemon.
Plugin will ensure that after stopping it, corresponding Docker container will be killed automatically.
Also, all daemon logs are redirected to separated file.

tasks {
    register("sftpServer") {
        doLast {
            aem.runDocker {
                operation("Runnning SFTP Server")
                image = "atmoz/sftp"
                command = "foo:pass:::upload"
                port(2222, 22)
            }
        }
    register("mockServer") {
        doLast {
            aem.runDocker {
                operation("Runnning Duckrails Mock Server")
                image = "iridakos/duckrails:release-v2.1.5"
                volume("duckrails", "/opt/duckrails/db")
                port(8080, 80)
            }
        }
    }
}

Properties expanding in an instance or package files

The properties syntax comes from Pebble Template Engine which means that all its features (if statements, for loops, filters, etc) can be used inside files being expanded.

Expanding properties could be used separately on any string or file source in any custom task by using the method common.prop.expand().