tfpdk
Terraform Provider Development Kit
DISCLAIMER: The author primarily works on the AzureRM Provider so this tool is heavily biased towards the code styles and requirements of that provider currently. Future development will look to genericise the output (eventually 😜)
NOTE: Expects to be run from the root of a validly named Terraform provider e.g. ./terraform-provider-myprovider
NOTE: Relies on a locally installed terraform binary and appropriately configured dev overrides file. This
allows the Terraform binary to skip the init
step by informing it where your compiled provider binary can be found. (you still currently need a local .tf file with your provider configured though, sorry!)
Run from the root of the provider project.
TODO
- untyped resource template
-
go fmt
outputs
- Add new resources / data sources to appropriate registration
- typed and untyped Data Sources
- init a new provider - git clone scaffold?
- optionally populate the Typed SDK into the init step?
- Dynamically read provider name for item generation
- Populate
IDValidationFunc()
in template for IDs (Pandora)
- Clients?
- Autocomplete CLI?
- Spike doc generation
- Extend doc gen to deal with Blocks
- migrate to
plugin sdk something else to read schema directly for doc gen as Terraform's JSON output lacks necessary detail. (e.g. timeouts, ForceNew flags etc)
- update commands and templates to allow non-hashicorp providers
and sources other than github (e.g. import paths)
Commands
Usage: tfpdk [--version] [--help] <command> [<args>]
Available commands are:
config Generate a local config file for common options.
datasource creates boiler-plate Data Sources.
document generates documentation from a resource.
init initialises a new provider based on the scaffold project.
resource creates boiler-plate resources.
servicepackage Creates a directory for a new Service Package and scaffolds out the basics to use it.
Usage Examples
Create a Typed Data Source for an existing Resource where the Resource's model is appropriate for use in the Data Source
tfpdk datasource -name ShinyNewService -servicepackage SomeCloudService -typed -useresourcemodel
Will create the following path {providername}/internal/services/SomeCloudService/shiny_new_service_data_source.go
Create an un-typed (traditional) resource
tfpdk resource -name ShinyNewService -servicepackage SomeCloudService
Will create the following path {providername}/internal/services/SomeCloudService/shiny_new_service_resource.go
and update registration.go
with the new resource in the SupportedResources()
func e.g.
func (r Registration) SupportedResources() map[string]*pluginsdk.Resource {
return map[string]*pluginsdk.Resource{
"azurerm_shiny_new_service": resourceShinyNewService(),
}
}
Create a typed updatable resource
tfpdk resource -name ShinyNewService -servicepackage SomeCloudService -has-update -typed
Will create the following path {providername}/internal/services/SomeCloudService/shiny_new_service_resource.go
and update registration.go
with the new resource in the Resources()
func, e.g.
func (r Registration) Resources() []sdk.Resource {
return []sdk.Resource{
ShinyNewServiceResource{},
}
}
Document your new resource (relies on the Description
fields in your schema, so populate those for best effect)
tfpdk document -type resource -name ShinyNewService -id "00000000-0000-0000-0000-000000000000"
Generate a Local config to set options and paths
tfpdk config
will create .tfpdk.hcl
in the root of the provider, which can be updated to configure provider specifics that may differ from the defaults. (by "defaults", I mean what we use on AzureRM), example output:
provider_name = "azurerm"
service_packages_path = "internal/services"
provider_github_org = "hashicorp"
docs_path = "docs"
resource_docs_directory_name = "r"
data_source_docs_directory_name = "d"
use_typed_sdk = false
Note: Command line options override values from the config file.
Command Documentation
tfpdk [command]