4.5 KiB

Raw Blame History

layout	page_title	sidebar_current	description
docs	Syntax - Configuration Language	configuration-syntax	HCL has its own syntax, intended to combine declarative structure with expressions in a way that is easy for humans to read and understand.

HCL Configuration Syntax

Other pages in this section have described various configuration constructs that can appear in HCL. This page describes the lower-level syntax of the language in more detail, revealing the building blocks that those constructs are built from.

This page describes the native syntax of HCL, which is a rich language designed to be easy for humans to read and write. The constructs in HCL can also be expressed in JSON syntax, which is harder for humans to read and edit but easier to generate and parse programmatically.

This low-level syntax of HCL is defined in terms of a syntax called HCL, which is also used by configuration languages in other applications, and in particular other HashiCorp products. It is not necessary to know all of the details of HCL in order to use Packer, and so this page summarizes the most important details. If you are interested, you can find a full definition of HCL syntax in the HCL native syntax specification.

Arguments and Blocks

HCL syntax is built around two key syntax constructs: arguments and blocks.

Arguments

An argument assigns a value to a particular name:

image_id = "abc123"

The identifier before the equals sign is the argument name, and the expression after the equals sign is the argument's value.

The context where the argument appears determines what value types are valid (for example, each source type has a schema that defines the types of its arguments), but many arguments accept arbitrary expressions, which allow the value to either be specified literally or generated from other values programmatically.

Blocks

A block is a container for other content:

source "amazon-ebs" "example" {
  ami_name = "abc123"

  tags {
    # ...
  }
}

A block has a type (source in this example). Each block type defines how many labels must follow the type keyword. The source block type expects two labels, which are amazon-ebs and example in the example above. A particular block type may have any number of required labels, or it may require none as with the nested tags block type.

After the block type keyword and any labels, the block body is delimited by the { and } characters. Within the block body, further arguments and blocks may be nested, creating a hierarchy of blocks and their associated arguments.

HCL uses a limited number of top-level block types, which are blocks that can appear outside of any other block in a configuration file. Most of Packer's features (including resources, input variables, output values, data sources, etc.) are implemented as top-level blocks.

Identifiers

Argument names, block type names, and the names of most Packer-specific constructs like resources, input variables, etc. are all identifiers.

Identifiers can contain letters, digits, underscores (_), and hyphens (-). The first character of an identifier must not be a digit, to avoid ambiguity with literal numbers.

For complete identifier rules, Packer implements the Unicode identifier syntax, extended to include the ASCII hyphen character -.

Comments

HCL supports three different syntaxes for comments:

# begins a single-line comment, ending at the end of the line.
// also begins a single-line comment, as an alternative to #.
/* and */ are start and end delimiters for a comment that might span over multiple lines.

The # single-line comment style is the default comment style and should be used in most cases. Automatic configuration formatting tools may automatically transform // comments into # comments, since the double-slash style is not idiomatic.

Character Encoding and Line Endings

Packer configuration files must always be UTF-8 encoded. While the delimiters of the language are all ASCII characters, Packer accepts non-ASCII characters in identifiers, comments, and string values.

Packer accepts configuration files with either Unix-style line endings (LF only) or Windows-style line endings (CR then LF), but the idiomatic style is to use the Unix convention, and so automatic configuration formatting tools may automatically transform CRLF endings to LF.

4.5 KiB Raw Blame History