#terraform #infrastructure #data-structures #serialization #serde

bin+lib tfschema-bindgen

Generate (de)serialization Rust code for Terraform Schema

4 releases

0.1.5 Feb 22, 2021
0.1.4 Dec 24, 2020
0.1.3 Sep 9, 2020

#264 in Encoding

MIT/Apache

51KB
996 lines

tfschema-bindgen

Build Status tfschena-bindgen on crates.io Documentation (latest release) License License

This crate aims to compile schemas extracted from Terraform providers into Serde type definitions.

Quick Start

A Terraform schema is required for generating Rust types responsible of deserialization and serialization. It can either be exported from your Terraform configuration or manually generated. We'll take the latter approach, therefore defining a reference schema with just one provider type having one attribute:

{
   "provider_schemas": {
       "test_provider": {
           "provider": {
               "version": 0,
               "block": {
                   "attributes": {
                       "base_url": {
                           "type": "string",
                           "description": "The url.",
                           "optional": true
                       }
                   }
               }
           }
       }
   },
   "format_version": "0.1"
}

In addition to a Rust library, this crate provides a binary tool tfbindgen to process Terraform schemas saved on disk. Outside of this repository, you may install the tool with:

cargo install tfschema-bindgen

Then use $HOME/.cargo/bin/tfbindgen.

We're going to use this tool assuming that we're inside the repository.

The following command will generate Serde bindings from the previous definitions, outputting those to test.rs module:

cargo run --bin tfbindgen -- test.json > test.rs

The following is a Rust example snippet comprising the previously generated bindings and a main function building on these in order deserialize a configuration descriptor adhering to our Terraform schema:

#![allow(unused_imports, non_snake_case, non_camel_case_types, non_upper_case_globals)]
use std::collections::BTreeMap as Map;
use serde::{Serialize, Deserialize};
use serde_bytes::ByteBuf as Bytes;

#[derive(Clone, Debug, PartialEq, PartialOrd, Serialize, Deserialize)]
pub struct config {
   pub data: Option<Vec<datasource_root>>,
   pub provider: Option<Vec<provider_root>>,
   pub resource: Option<Vec<resource_root>>,
}

#[derive(Clone, Debug, PartialEq, PartialOrd, Serialize, Deserialize)]
pub enum datasource_root {
}

#[derive(Clone, Debug, PartialEq, PartialOrd, Serialize, Deserialize)]
pub enum provider_root {
   test_provider(Vec<test_provider_details>),
}

#[derive(Clone, Debug, PartialEq, PartialOrd, Serialize, Deserialize)]
pub enum resource_root {
}

#[derive(Clone, Debug, PartialEq, PartialOrd, Serialize, Deserialize)]
pub struct test_provider_details {
   pub base_url: Option<String>,
}

const TF_JSON_CONFIG: &str = r#"{
   "provider": [
     {
       "test_provider": [
         {
           "base_url": "https://acme.com/foo"
         }
       ]
     }
   ]
}"#;

fn main() -> Result<(), std::io::Error> {
   let res: config = serde_json::from_str(TF_JSON_CONFIG).unwrap();

   assert_eq!(res.provider.as_ref().map(|x| x.is_empty()), Some(false));
   assert_eq!(
       res.provider.as_ref().map(|x| x.get(0).is_none()),
       Some(false)
   );
   let prv = res
       .provider
       .as_ref()
       .and_then(|x| x.get(0))
       .and_then(|x| match x {
           provider_root::test_provider(p) => p.get(0),
       });
   assert_eq!(prv.is_none(), false);
   assert_eq!(
       prv.and_then(|x| x.base_url.to_owned()),
       Some("https://acme.com/foo".to_owned())
   );
   print!("success!\n");
   Ok(())
}

Quickstart Example

In addition to a Rust library and generation tool, this crate provides the above example which can be executed using the following command:

cargo run --example quickstart

Consuming third-party Terraform schemas

In order to operate on Terraform configuration descriptors of third-party providers, Rust bindings have to be generated using the provided schema descriptor in the JSON format.

Firstly, create a minimal Terraform configuration declaring the target provider. The following is an example for enabling the Amazon Web Services (AWS) Terraform provider:

provider "aws" {
 version = ">= 2.31.0, < 3.0"
}

Initialize Terraform so that configured providers are installed in the local environment:

terraform init

Secondly, extract the schema for the providers defined in the Terraform configuration, AWS in this case:

terraform providers schema -json > aws-provider-schema.json

Finally, generate the Rust (de)serialization types for the given provider using the following command (assuming you are inside the repository):

cargo run --bin tfbindgen -- aws-provider-schema.json > aws_provider_schema.rs

In order do (de)serialize provider's configuration, import the generated module in your application.

License

This project is available under the terms of either the Apache 2.0 license or the MIT license.

Dependencies

~4.5MB
~93K SLoC

-4`