tf-output

Fetches terraform outputs and lets you:

• print them to stdout in different formats
• call another command with them exposed in the environment
• export them to your current shell

Getting Started

You need to install this module and then run it's command from somewhere that has a property initialized terraform environment.

Install

npm install -g tf-output

Usage

Assuming you have a module called database.tf with an output called DATABASE_URL in a folder called terraform/database and you have already set up terraform...

Examples

tf-output database

Would print out something like:

DATABASE_URL="http://database.totallysecure.com"

And:

tf-output database -- node src/app.js

Would call node src/app.js with the DATABASE_URL available in the process environment.

You can also

export $(tf-output database)

Which will populate outputs in to your current shell.

Configuration

Module

-m or --module specifies which module to obtain output from. -m alone uses the dir name as the value. So

tf-output api -m

calls terraform output -m api in terraform/api

Output Format

-f or --format allows you to specify an alternate output format, currently json is supported. This flag is ignored if a command is specified.

Backend Initialization

-a or --auto-init allows you to auto-initialize terraform (terraform init) prior to copying outputs.

-g or --auto-init-get will pull down modules mentioned in the root module during auto-initialization.

-i or --init-opts allows you to pass additional options to terraform init (e.g. -backend-config).

Output Flattening

-fl or --flatten will flatten output values that are objects by concatenating key names.

-fd or --flatten-delimiter allows you to specify the delimiter to use while concatenating key names (_ by default).

Advanced Configuration

Plan Check

-c or --check-plan checks if terraform plan has any unapplied changes (and aborts if it does).

-o or --plan-opts allows you to pass additional options to terraform plan (e.g. -var-file).

Path Template

-p or -path specifies a path template. tf-output looks for modules according to a path template, which by default is terraform/{dir} - so tf-output api database would look for a module in two directories:

• terraform/database
• terraform/api

You can use more complex path templates. Imagine you deploy your app using the following command:

deploy --stage=dev --region=us-east-1

You might organize your terraform definitions in a number of ways and you can customize where tf-output looks to suit your needs.

tf-output database -p {stage}/{region}/terraform/{dir}

This command would fail, because while it knows the module you are trying to load, it doesn't know what stage and region should be.

You can specify them:

tf-output database api -p {stage}/{region}/terraform/{dir} --stage=dev --region=us-east-1

This would cause tf-output to load modules from:

• dev/us-east-1/terraform/api
• dev/us-east-1/terraform/database

If you run a command through tf-output, it will look ahead for substitution values. This works the same:

tf-output database api -p {stage}/{region}/terraform/{dir} -- deploy --stage=dev --region=us-east-1

Except instead of printing the outputs out to stdout, it would exec deploy --stage=dev --region=us-east-1 with environment variables set, and the path template would still be substituted with the right stage and region. This saves you havin to repeat region, stage, etc.

.tfoutput

In many cases the arguments you will specify to tf-output are always going to be the same, for example the -p path template argument. You can put these in a .tfoutput file which will be read on every execution.

Example .tfoutput file:

{
    "path": "terraform/{dir}/{stage}/{region}",
    "module": true,
    "auto-init": true,
    "auto-init-get": true
}