Home

Awesome

Terraform graph beautifier

Command line tool allowing to convert the barely usable output of the terraform graph command to something more meaningful and explanatory.

terraform graph raw outputProcessed output
??

Installation

You can download the binary corresponding to the required platform from the release page.

You can also build the binary from the sources:

go install github.com/pcasteran/terraform-graph-beautifier

Finally, you can also use the Docker images available for the Linux and Darwin operating systems:

# Linux
docker pull ghcr.io/pcasteran/terraform-graph-beautifier:latest-linux

# Darwin
docker pull ghcr.io/pcasteran/terraform-graph-beautifier:latest-darwin

Basic usage

cd samples/config1/
terraform init

# Using the binary.
terraform graph | terraform-graph-beautifier \
    --output-type=cyto-html \
    > config1.html

# Using the Docker image.
terraform graph | docker run --rm -i \
    --name terraform-graph-beautifier \
    ghcr.io/pcasteran/terraform-graph-beautifier:latest-linux \
    --output-type=cyto-html \
    > config1.html

Input and outputs

The command performs the following:

  1. Parsing of the Graphviz Dot script generated by terraform graph from the standard input or the specified file (--input parameter).
  2. Extraction of the graph containing the Terraform configuration elements and their dependencies.
  3. Generation of the result to the standard output or file (--output parameter). The following output types (--output-type parameter) are supported:
    • cyto-html (default): an HTML page using Cytoscape.js to render the graph ( see example);
    • cyto-json: a JSON document of the graph in the Cytoscape.js format (see example);
    • graphviz: a cleaned and prettier Dot script that can be piped to a Graphviz rendering command (see example).

Input graph loading and processing

The loading of the input graph involves the following steps:

  1. Cleaning the Dot script
    1. Renaming the nodes using a more consistent pattern: [root] rsc_type.rsc_name => module.root.rsc_type.rsc_name.
    2. Using the ' character instead of " fot the maps keys: buckets["artefacts"] => buckets['artefacts'].
    3. Removing the nodes and edges generated by Terraform but not corresponding to configuration elements (aka TF junk). This can be deactivated via the --keep-tf-junk parameter.
  2. Filtering
    • Using user-provided pattern(s) to exclude some elements (resource, var, module, provider, ...) from the output.
    • These patterns are Go regexp and are matched line by line against the output of the cleaning step, so use the "root.rsc_type.rsc_name" naming convention.
    • These patterns are provided using the --exclude parameter, you can repeat it multiple times.

Output configuration

Modules embedding

An important option is to choose whether to embed a submodule in its parent module or not. The --embed-modules parameter allows to control this behavior:

As a rule of thumb, --embed-modules=true works well for small to medium size graphs but for larger ones it can produce a very dense and compact result with a lot of overlapping nodes and intersecting edges.

Output type--embed-modules=true--embed-modules=false
cyto-html??
graphviz??

HTML output templating

The command uses Go templates to create the HTML output. The following annotations will be replaced during the output generation:

A basic template, with sensible default values for the graph rendering (style and layout), is provided and is embedded in the binary.

If you want to customize the output, you can provide your own template and specify to use it with the --cyto-html-template parameter. It is good practice to check in your custom templates alongside your Terraform configuration.