Marketplace Autogen is a Java-based tool for generating fully functional Terraform modules or Deployment Manager templates from simplified YAML, JSON, or Prototext configuration files. It currently supports single- and multi-VM configurations.
This tool also generates the UI metadata used to actuate the Marketplace deployment UI. For Terraform, the UI metadata is structured as BlueprintMetadata.
With AutogenCli, you can generate a solution package through the command line, by providing your configuration in a JSON, YAML, or Prototext file.
If you want to generate a deployment package for a single solution, your configuration file should follow the schema described in the DeploymentPackageInput proto. If you want to generate deployment packages for multiple solutions, your configuration file should follow the schema described in the BatchInput proto. Both specs can be found in autogen.proto.
To review these options at any time, provide --help as the argument for AutogenCli.
--single_input
This parameter is used to indicate that Autogen will be run on a configuration that contains a DeploymentPackageInput spec. If an argument is specified, AutogenCli will interpret it as the path to the file to read the spec from; if it is empty, AutogenCli will read from stdin. Either --single_input or --batch_input must be provided as a parameter.
--batch_input
This parameter is used to indicate that Autogen will be run on a configuration that contains a BatchInput spec. If an argument is specified, AutogenCli will interpret it as the path to the file to read the spec from; if it is empty, AutogenCli will read from stdin. Either --single_input or --batch_input must be provided as a parameter.
--input_type (optional, defaults to PROTOTEXT)
This indicates the format of the spec that AutogenCli will read in. The available options for this parameter are: YAML, JSON, PROTOTEXT, and WIRE (binary prototext).
--output_type (optional, defaults to PROTOTEXT)
This indicates the format of the spec that AutogenCli will write out. The available options for this parameter are: YAML, JSON, PROTOTEXT, WIRE, and PACKAGE.
--output (optional)
If the --output_type is PACKAGE, then the argument of this parameter will be interpreted as a destination folder for the deployment files to be written to. If it is left empty or nothing is provided, Autogen will use the current directory iun which the binary is being run.
For all other --output_type options, the argument of this parameter will be interpreted as a path to where the file should be output to. If it is left empty or nothing is provided, Autogen will default to stdout.
--exclude_shared_support_files (optional)
If this parameter is provided, Autogen will NOT include the shared support files used for deployment in this solution (look here for those files). By default, Autogen always includes these files, which is the recommended option. This is only relevant for Deployment Manager.
We have provided a full featured example configuration in the example-config folder.
Please refer to the comments inside the example configuration and the proto files (autogen.proto, deployment_package_autogen_spec.proto and marketing_info.proto) for more information about their fields, including which of them are required and which are optional.
To create your deployment package, change the fields of the provided examples to reflect the configuration of your solution, and then run Autogen.
It's easiest to use our released docker images. If you want to build and run the tool from source, see the Development section below.
Make sure that docker has been installed.
Set up the following alias, to keep your commands clean:1
alias autogen='docker run \
--rm \
--workdir /mounted \
--mount type=bind,source="$(pwd)",target=/mounted \
--user $(id -u):$(id -g) \
gcr.io/cloud-marketplace-tools/dm/autogen'
autogen --help
The command below creates a deployment package according to the spec in example-config/solution.yaml, and then outputs all of the resulting files into solution_folder:
mkdir solution_folder
autogen \
--input_type YAML \
--single_input example-config/solution.yaml \
--output_type PACKAGE \
--output solution_folder
After the command has completed, you can then compress the solution_folder folder, and use Partner Portal to upload it as your deployment package.
NOTE: Only files in the current working folder can be seen and manipulated by the tool, so you must make sure to pass in the input and output file parameters using relative paths.
Other sample configurations are also available in the testdata folder.
The command below runs a sample test proto definition transformation, outputting the deployment package in the JSON format:
autogen \
--input_type PROTOTEXT \
--single_input javatests/com/google/cloud/deploymentmanager/autogen/testdata/singlevm/full_features/input.prototext \
--output_type JSON
1A few notes about the alias:
--mounting the current working directory. Within the container runtime, the content is available at /mounted.--workdir to /mounted for the running container, so that file parameters like --single_input can use relative paths.--user to instruct docker to run as the current user, instead of root.To build all of the artifacts, run the following bazel command:
bazel build java/com/google/cloud/deploymentmanager/autogen/...:all
If you want to, you can also build AutogenCli_deploy.jar target, to run the jar as a standalone application:
bazel build java/com/google/cloud/deploymentmanager/autogen/cli:AutogenCli_deploy.jar
An example:
bazel-bin/java/com/google/cloud/deploymentmanager/autogen/cli/AutogenCli \
--input_type YAML \
--single_input example-config/solution.yaml \
--output_type PACKAGE \
--output solution_folder
bazel test javatests/com/google/cloud/deploymentmanager/autogen:all
If you see errors when trying to build Autogen using bazel, try to run the following command, and then try again:
bazel clean --expunge