Python binding for Terraform.

$ pip install libterraform

NOTE

• Please install version 0.3.1 or above, which solves the memory leak problem.
• This library does not support multithreading.

TerraformCommand is used to invoke various Terraform commands.

Now, support all commands (plan, apply, destroy etc.), and return a CommandResult object. The CommandResult object has the following properties:

• retcode indicates the command return code. A value of 0 or 2 is normal, otherwise is abnormal.
• value represents command output. If json=True is specified when executing the command, the output will be loaded as json.
• json indicates whether to load the output as json.
• error indicates command error output.

To get Terraform verison:

>>> from libterraform import TerraformCommand
>>> TerraformCommand().version()
<CommandResult retcode=0 json=True>
>>> _.value
{'terraform_version': '1.2.2', 'platform': 'darwin_arm64', 'provider_selections': {}, 'terraform_outdated': False}
>>> TerraformCommand().version(json=False)
<CommandResult retcode=0 json=False>
>>> _.value
'Terraform v1.2.2\non darwin_arm64\n'

To init and apply according to Terraform configuration files in the specified directory:

>>> from libterraform import TerraformCommand
>>> cli = TerraformCommand('your_terraform_configuration_directory')
>>> cli.init()
<CommandResult retcode=0 json=False>
>>> cli.apply()
<CommandResult retcode=0 json=True>

Additionally, run() can execute arbitrary commands, returning a tuple (retcode, stdout, stderr).

>>> TerraformCommand.run('version')
(0, 'Terraform v1.2.2\non darwin_arm64\n', '')
>>> TerraformCommand.run('invalid')
(1, '', 'Terraform has no command named "invalid".\n\nTo see all of Terraform\'s top-level commands, run:\n  terraform -help\n\n')

TerraformConfig is used to parse Terraform config files.

For now, only supply TerraformConfig.load_config_dir method which reads the .tf and .tf.json files in the given directory as config files and then combines these files into a single Module. This method returns (mod, diags) which are both dict, corresponding to the *Module and hcl.Diagnostic structures in Terraform respectively.

>>> from libterraform import TerraformConfig
>>> mod, _ = TerraformConfig.load_config_dir('your_terraform_configuration_directory')
>>> mod['ManagedResources'].keys()
dict_keys(['time_sleep.wait1', 'time_sleep.wait2'])

If you want to develop this library, should first prepare the following environments:

• GoLang (Version 1.18+)
• Python (Version 3.7~3.11)
• GCC

Then, initialize git submodule:

$ git submodule init
$ git submodule update

pip install necessary tools:

$ pip install poetry pytest

Now, we can build and test:

$ poetry build -f wheel
$ pytest

Terraform is a great tool for deploying resources. If you need to call the Terraform command in the Python program for deployment, a new process needs to be created to execute the Terraform command on the system. A typical example of this is the python-terraform library. Doing so has the following problems:

• Requires Terraform commands on the system.
• The overhead of starting a new process is relatively high.

This library compiles Terraform as a dynamic link library in advance, and then loads it for calling. So there is no need to install Terraform, nor to start a new process.

In addition, since the Terraform dynamic link library is loaded, this library can further call Terraform's internal capabilities, such as parsing Terraform config files.