HCL (HashiCorp Configuration Language) is a configuration language built by HashiCorp. The goal of HCL is to build a structured configuration language that is both human and machine friendly for use with command-line tools, but specifically targeted towards DevOps tools, servers, etc.
NOTE: This is major version 1 of HCL, which is now in maintenence mode only. We strongly recommend that new projects use HCL 2 instead. For more information, see our version selection guide.
HCL is also fully JSON compatible. That is, JSON can be used as a completely valid input to a system expecting HCL. This helps to make systems interoperable with other systems.
HCL is heavily inspired by libucl, nginx configuration, and others similar.
A common question when viewing HCL is to ask the question: why not JSON, YAML, etc.?
Prior to HCL, the tools we built at HashiCorp used a variety of configuration languages from full programming languages such as Ruby to complete data structure languages such as JSON. What we learned is that some people wanted human-friendly configuration languages and some people wanted machine-friendly languages.
JSON fits a nice balance in this, but is fairly verbose and most importantly doesn't support comments. With YAML, we found that beginners had a really hard time determining what the actual structure was, and ended up guessing more often than not whether to use a hyphen, colon, etc. in order to represent some configuration key.
Full programming languages such as Ruby enable complex behavior a configuration language shouldn't usually allow, and also forces people to learn some set of Ruby.
Because of this, we decided to create our own configuration language that is JSON-compatible. Our configuration language (HCL) is designed to be written and modified by humans. The API for HCL allows JSON as an input so that it is also machine-friendly (machines can generate JSON instead of trying to generate HCL).
Our goal with HCL is not to alienate other configuration languages. It is instead to provide HCL as a specialized language for our tools, and JSON as the interoperability layer.
For a complete grammar, please see the parser itself. A high-level overview of the syntax and grammar is listed here.
-
Single line comments start with
#
or//
-
Multi-line comments are wrapped in
/*
and*/
. Nested block comments are not allowed. A multi-line comment (also known as a block comment) terminates at the first*/
found. -
Values are assigned with the syntax
key = value
(whitespace doesn't matter). The value can be any primitive: a string, number, boolean, object, or list. -
Strings are double-quoted and can contain any UTF-8 characters. Example:
"Hello, World"
-
Multi-line strings start with
<<EOF
at the end of a line, and end withEOF
on its own line (here documents). Any text may be used in place ofEOF
. Example:
<<FOO
hello
world
FOO
-
Numbers are assumed to be base 10. If you prefix a number with 0x, it is treated as a hexadecimal. If it is prefixed with 0, it is treated as an octal. Numbers can be in scientific notation: "1e10".
-
Boolean values:
true
,false
-
Arrays can be made by wrapping it in
[]
. Example:["foo", "bar", 42]
. Arrays can contain primitives, other arrays, and objects. As an alternative, lists of objects can be created with repeated blocks, using this structure:service { key = "value" } service { key = "value" }
Objects and nested objects are created using the structure shown below:
variable "ami" {
description = "the AMI to use"
}
This would be equivalent to the following json:
{
"variable": {
"ami": {
"description": "the AMI to use"
}
}
}
Thanks to: