This is a toolset package for working with AEM and especially the Java Content Repository (JCR) via command line tools. It tries to utilize the unix philosophy by reading and writing plaintext in order to interoperate with common tools such as grep, cut, sed and awk. Mostly it depends on the Sling json apis for interfacing.

A quick install script is provided for convenience:

$ curl https://raw.githubusercontent.com/bjorns/aem-cmd/master/get-acmd.sh | bash

Note: The install script requires sudo access. There is always a risk to running arbitrary code off the internet like this.

For manual install, acmd is available in PyPI. To install, simply call pip like any other python package.

$ pip install aem-cmd
...
$ acmd
Usage: acmd [options] <tool> <args>
Run 'acmd help' for list of available tools

Options:
  -h, --help            show this help message and exit
  -s str, --server=str  server name

Acmd comes with a bash completion script but for technical reason it cannot be installed by pip so to install it use the install_bash_completion command.

$ sudo acmd install_bash_completion
Password:
Installed bash completion script in /etc/bash_completion.d

Note 1: The bash completion works best in bash 4. Due to licensing issues Mac OS still comes with bash 3 by default but it is possible to upgrade. See here.

Note 2: The quick install scripts automatically adds the bash completion.

Acmd is built around tools. Each tool represents a resource in the system. Packages, bundles and users each have separate tools for operating on them. Think of it in terms of REST services. Primarily you interact with a resource, not an action.

The help tool lists all installed tools

$ acmd help
Available tools:
  inspect
  bundles
  help
  packages

To get more information on available actions call help on that tool:

$ acmd help packages
Usage: acmd packages [options] [list|build|install|upload|download] [<zip>|<package>]

Options:
  -h, --help            show this help message and exit
  -v VERSION, --version=VERSION
                        specify explicit version
  -g GROUP, --group=GROUP
                        specify explicit group
  -r, --raw             output raw response data
  -c, --compact         output only package name
  -i, --install         install package after upload

The JCR tools are considered so common they are not grouped together but come as separate commands.

$ acmd ls /
index.servlet
bundles
rep:policy
services
home
....

$ acmd find /content/catalog
...

$ acmd cat /content
jcr:primaryType:	sling:OrderedFolder
jcr:createdBy:	admin
jcr:mixinTypes:	[u'sling:Redirect', u'mix:lockable', u'rep:AccessControllable']
sling:target:	/geohome
sling:resourceType:	sling:redirect
jcr:created:	Fri Jun 13 2014 14:17:56 GMT-0400

$ acmd setprop prop=cheese /content/catalog/product4711
/content/catalog/product4711
$ acmd cat /content/catalog/product4711
serial_nbr: 1234
...
prop:   cheese

Multiple properties can be set comma separated. Just quote the values if there are commas or spaces in the values. Quoting will also force the type to string, otherwise rudimentary type inference is performed recognizing numbers and booleans.

$ acmd setprop prop1="I like cheese",prop2="I also like wine" /content/catalog/product4711

The setprop tool also takes paths on stdin. The following line sets the property on all nodes under /content/catalog

$ acmd find /content/catalog | acmd setprop prop="I like cheese"

Note that the setprop command will create the path if it does not exist. The following will create a new folder:

$ acmd setprop jcr:primaryType=sling:Folder /content/new_folder

$ acmd rmprop prop /content/catalog/product4711
/content/catalog/product4711

Works very similar to setprop, takes multiple paths on stdin and can take multiple comma separated property names as prop0,prop1,prop2.

The search tool uses the AEM query API to find nodes which contain a given property value.

$ acmd search serial_nbr=1234
/content/catalog/product4711

$ acmd rm /content/catalog/product4711
/content/catalog/product4711

The rm tool if given no argument will read node paths from standard input.

$ acmd find /content/catalog | grep product | acmd rm
....

The packages tool supports up- and downloading, installing and uninstalling packages. If you install the bash completion script, package names will be autocompleted. In addition to autocomplete the tool also automatically finds group and version of the latest package so only the simple package name needs to be supplied as argument.

By default the packages tool lists all installed packages.

$ acmd packages list
...
day/cq540/product   cq-portlet-director 5.4.38
day/cq550/product   cq-upgrade-acl  5.5.2
day/cq560/collab/blog   cq-collab-blog-pkg  5.6.17
day/cq560/collab/calendar   cq-collab-calendar-pkg  5.6.12
day/cq560/social/calendar   cq-social-calendar-pkg  1.0.28
....

The packages commands require only that you provide the name of the package. The group and the latest version will be located automatically. If there are overlaying grops or you want a specific version you may specify them using the -g and -v flags.

$ acmd packages build --raw cq-upgrade-acl
{"success":true,"msg":"Package built"}

$ acmd packages install --raw cq-upgrade-acl
{"success":true,"msg":"Package installed"}

You may install a properly generated package zip e.g. downloaded from another instance.

$ acmd packages upload new-catalog-1.0.zip

$ acmd users list
...

$ acmd users create --password=foobar jdoe
/home/users/j/jdoe

$ acmd users setprop age=29,name="John Doe" jdoe
/home/users/j/jdoe

$ acmd groups list
....

The list action is the default action of the groups tool so acmd groups will suffice.

$ acmd groups create editors
/home/groups/e/editors

$ acmd groups adduser editors jdoe
/home/groups/e/editors

The bundles tool can list, start and stop jackrabbit OSGi bundles. If you install the bash completion script bundle names will be autocompleted. Like the packages tool, the group and version of the bundle will be inferred for all commands.

List all bundles in the system.

$ acmd bundles list
...
org.apache.sling.extensions.webconsolesecurityprovider  1.0.0   Active
org.apache.sling.jcr.jcr-wrapper    2.0.0   Active
org.apache.tika.core    1.3.0.r1436209  Active
org.apache.tika.parsers 1.3.0.r1436209  Active
biz.aQute.bndlib    1.43.0  Active
com.day.cq.collab.cq-collab-blog    5.6.10  Active
com.day.cq.collab.cq-collab-calendar    5.6.9   Active
...

The output is tab-separated to make it easy to read specific data. E.g. to get the version of the bndlib bundle:

$ acmd bundles list | grep bndlib | cut -f 2
1.43.0

$ acmd bundles stop biz.aQute.bndlib
$

All bundles commands support the --raw flag which prints raw json output.

$ acmd bundles start biz.aQute.bndlib --raw
{
    "fragment": false,
    "stateRaw": 32
}
$

Note: the groovy tool requires the cq-groovy-console bundle from Citytech.

The groovy tool allows execution of arbitrary code server side. Put your code in a file and execute with:

$ acmd groovy list-entire-catalog.groovy
....

For more documentation on how to write groovy scripts install the cq-groovy-console bundle and go to http://localhost:4502/etc/groovyconsole.html

The assets tool controls interactions with the AEM dam.

The import command allows for importing single files and entire directories into the DAM.

To import a single file into /content/dam

$ acmd assets import ~/Pictures/selfie.jpg

Send file to a separate folder

$ acmd assets import -d /content/dam/selfies ~/Pictures/selfie.jpg

If a directory is given it will be recreated at /content/dam. The following will create the node /content/dam/selfies.

$ acmd assets import ~/Images/selfies

The tool maintains a set of lockfiles under /tmp/acmd_assets_ingest on the local filesystem where each uploaded file is marked so that aborted imports can be resumed without redoing previous imports. The lockfile directory can be overridden with the --lock-dir flag.

$ acmd assets import -l ~/my_lock_dir ~/Images/selfies

The import tool allows the user to filter imported files by path or file type. The following is an example filter config file.

{
    "accept": {
        "filetypes": [
            "png",
            "jpg",
            "jpeg",
            "indd",
            "ai"
        ],
        "paths": [
            "Images/good_ones",
            "Images/average_ones"
        ]
    },
    "reject": {
        "filetypes": ["DS_Store"],
        "paths": [
            "Images/good_ones/exception"
        ]
    }
}

If the file is saved as filter_config.json it can be applied by

$ acmd assets import -f filter_config.json ~/Images/selfies

The filtering process is simple:

• If the 'rejected' entry is not defined, only accepted entries are imported
• If the 'rejected' entry is defined, all files are imported except the ones that match the rejected

Note: the rcp tool requires the JackRabbit Vault rcp bundle

The rcp tool allows the user to copy large repositories of data between two aem instances without large packages or configuring replication.

Rcp is implemented as a fetch operation. You call the destination server, give it the url to the source data and it will start a job to fetch the data.

$ acmd -s uat rcp fetch --source-host prod-author:4502 /content/dam/mysite

An optional second argument defines the target path. If omitted, destination is the same as source

$ acmd -s uat rcp fetch --source-host prod-author:4502 /content/dam/mysite /content/dam/mysite_backup

The fetch operation is a synchronous macro operation. For very large operations you can also utilize the async low level commands but creating a task and managing it on your own. The create operation prints the newly created task id. This id is then used as a reference when managing the task.

$ acmd rcp create --source-host prod-author:4502 /content/dam/mysite
rcp-45cfa2
$ acmd rcp ls
rcp-45cfa2  NEW     /content/dam/mysite
$ acmd rcp start rcp-45cfa2
$ acmd rcp ls
rcp-45cfa2  RUNNING     /content/dam/mysite
...
$ acmd rcp ls
rcp-45cfa2  ENDED     /content/dam/mysite
$ acmd rcp rm rcp-45cfa2

By default the task copies all nodes that don't already exist on the destination or have a lastModified date later on the source than on the destination. This can be controlled using the -i and the -n flags.

The backend storage tool can trigger optimization jobs.

$ acmd storage optimize

Asynchronously triggers a tar optimization job for the filesystem backend.

$ acmd storage gc

Asynchronously triggers a garbage collection job.

Acmd config is specified in ~/.acmd.rc If this file does not exist at first execution it will be created from a template. Here is what the template file looks like:

# This is a template configuration file for the AEM Command line tools.
# acmd expects this file to exist as ~/.acmd.rc

# Each server is defined by a section named [server <name>].
# This server can then be accessed via the --server option.
[server localhost]
host=http://localhost:4502
username=admin
password=admin

# The default server option specifies which server to use if
# the --server option is not specified.
[settings]
default_server=localhost

# Add custom tools directories here as
# prefix=<custom_tool_dir>
[projects]

The most common use of the config file is to add additional servers. You will want to add a server entry for each instance in your setup. So one for your local, one for integration testing, one staging server and one each for author and publish instances in production.

Once this is done you can easily run any command on any server via the -s flag. Like for example

$ acmd -s prod-author bundles
....

The projects section enables you to add project specific tools you your arsenal. Let's say you have created a custom tool for handling a product catalog under ~/my_project/acmd-tools/catalog.py. Now what you do is add the following to your .acmd.rc

[projects]
custom=~/my_project/acmd-tools

All python files in aem-tools will be imported and your tool should show up in with the help command as custom:catalog

$ acmd help
 ...
custom:catalog

Writing a custom tool is relatively easy. All of acmd is written in Python and relies on the requests http client library for communication with AEM. The best way to learn how to extend is to read the tools code under acmd/tools.

Nevertheless here is boilerplate for creating a new tool.

# coding: utf-8
from acmd.tool import tool

@tool('catalog', ['inspect', 'upload'])
class CatalogTool(object):
    def execute(self, server, argv):
        sys.stdout.write("Hello, world.\n")

There are two essential parts here. The @tool decorator takes care of declaring the tool so it can be found by acmd. The first argument is the label used for the tool on the command line. Note that for custom tools declared under [projects] a prefix will be added to the tool name. The second argument is a list of commands available under the tool. Remember that under normal circumstances the tool represents a resource and the first argument is a command to perform on that resource. This list is used for autocompletion and can be omitted for simpler tools.

The execute() method takes a server argument containing all info on the currently selected server and argv is a list of tool arguments starting with the tool name. See for example the bundles tool for info on how to use the optparse package to add tool specific options.

And that is pretty much it. The tool does not have to have any specific name (though a -Tool suffix is idiomatic), and it does not have to inherit any specific class.

• The inspiration for acmd comes from the Polopoly CMS command line tools - https://github.com/polopolyps/pcmd
• The AEM Curl command reference was incredibly helpful during the development of acmd - https://gist.github.com/sergeimuller/2916697
• oak-run is a another great command line tool for maintaining AEM instances - https://github.com/apache/jackrabbit-oak/tree/trunk/oak-run