Below is a list of linters supported by haml-lint
, ordered alphabetically.
- ClassAttributeWithStaticValue
- ClassesBeforeIds
- ConsecutiveComments
- ConsecutiveSilentScripts
- EmptyScript
- ImplicitDiv
- HtmlAttributes
- LeadingCommentSpace
- LineLength
- MultilinePipe
- ObjectReferenceAttributes
- RuboCop
- RubyComments
- SpaceBeforeScript
- SpaceInsideHashAttributes
- TagName
- TrailingWhitespace
- UnnecessaryInterpolation
- UnnecessaryStringOutput
Prefer static class attributes over hash attributes with static values.
Bad
%tag{ class: 'my-class' }
Good
%tag.my-class
Unless you are assigning a dynamic value to the class attribute, it is terser to use the inline tag syntax to specify the class(es) an element should be assigned.
List classes before IDs in tags.
Bad: ID before class
%tag#id.class
Good
%tag.class#id
These attributes should be listed in order of their specificity. Since the tag name (if specified) always comes first and has the lowest specificity, classes and then IDs should follow.
Consecutive comments should be condensed into a single multiline comment.
Bad
-# A collection
-# of many
-# consecutive comments
Good
-#
A multiline comment
is much more clean
and concise
Option | Description |
---|---|
max_consecutive |
Maximum number of consecutive scripts allowed before warning (default 2 ) |
Avoid writing multiple lines of Ruby using silent script markers (-
).
Bad
- expression_one
- expression_two
- expression_three
Better
:ruby
expression_one
expression_two
expression_three
In general, large blocks of Ruby code in HAML templates are a smell, and this
check serves to warn you of that. However, for the cases where having the code
inline can improve readability, you can signal your intention by using a
:ruby
filter block instead.
Don't write empty scripts.
Bad: script marker with no code
-
Good
- some_expression
These serve no purpose and are usually left behind by mistake.
Don't use the HTML-style attributes syntax to define attributes for an element.
Bad
%tag(lang=en)
Good
%tag{ lang: 'en' }
While the HTML-style attributes syntax can be terser, it introduces additional complexity to your templates as there are now two different ways to define attributes. Standardizing on when to use HTML-style versus hash-style adds greater cognitive load when writing templates. Using one style makes this easier.
Avoid writing %div
when it would otherwise be implicit.
Bad: div
is unnecessary when class/ID is specified
%div.button
Good: div
is required when no class/ID is specified
%div
Good
.button
HAML was designed to be concise, and not embracing this philosophy makes the tool less useful.
Separate comments from the leading #
by a space.
Bad
-#Comment with no space
Good
-# Comment with space
The latter is more readable.
Option | Description |
---|---|
max |
Maximum number of columns a single line can have. (default 80 ) |
Wrap lines at 80 characters. You can configure this amount via the max
option on the linter, e.g. by adding the following to your .haml-lint.yml
:
linters:
LineLength:
max: 100
Long lines are harder to read and usually indicative of complexity. You can avoid them by splitting long attribute hashes on a comma, for example:
%tag{ attr1: 1,
attr2: 2,
attr3: 3 }
This significantly improves readability.
Don't span multiple lines using the multiline pipe (|
) syntax.
Bad
%p= 'Some' + |
'long' + |
'string' |
Good: use helpers to generate long dynamic strings
%p= generate_long_string
Good: split long method calls on commas
%p= some_helper_method(some_value,
another_value,
yet_another_value)
Good: split attribute definitions/hashes on commas
%p{ data: { value: value,
name: name } }
The multiline bar was
made awkward intentionally.
haml-lint
takes this a step further by discouraging its use entirely, as it
almost always suggests an unnecessarily complicated template that should have
its logic extracted into a helper.
Don't use the object reference syntax to set the class/id of an element.
Bad
%li[@user]
= @user.name
Good
%li.user{ id: "user_#{@user.id}" }
= @user.name
The object reference syntax is a bit magical, and makes it difficult to find where in your code a particular class attribute is defined. It is also tied directly to the class names of the objects you pass to it, creating an unnecessarily coupling which can make refactoring your models affect your views.
Option | Description |
---|---|
ignored_cops |
Array of RuboCop cops to ignore. |
This linter integrates with RuboCop
(a static code analyzer and style enforcer) to check the actual Ruby code in
your templates. It will respect any RuboCop-specific configuration you have
set in .rubocop.yml
files, but will explicitly ignore some checks that
don't make sense in the context of HAML documents (like
Style/BlockAlignment
).
-# example.haml
- name = 'James Brown'
- unused_variable = 42
%p Hello #{name}!
Output from haml-lint
example.haml:3 [W] Useless assignment to variable - unused_variable
You can customize which RuboCop warnings you want to ignore by modifying
the ignored_cops
option (see config/default.yml
for the full list of ignored cops).
Prefer HAML's built-in comment over ad hoc comments in Ruby code.
Bad: Space after #
means comment is actually treated as Ruby code
- # A Ruby comment
Good
-# A HAML comment
While both comment types will result in nothing being output, HAML comments are a little more flexible in that you can have them span multiple lines, e.g.
-# This is a multi-line
HAML comment
Separate Ruby script indicators (-
/=
) from their code with a single space.
Bad: no space between =
and some_expression
=some_expression
Good
= some_expression
Good
- some_value = 'Hello World'
Ensuring space after -
/=
enforces a consistency that all HAML tags/script
indicators are separated from their inline content by a space. Since it is
optional to add a space after -
/=
but required when writing %tag
or
similar, the consistency is best enforced via a linter.
Check the style of hash attributes against one of two possible preferred
styles, space
(default) or no_space
:
Bad: inconsistent spacing inside hash attributes braces
%tag{ foo: bar}
%tag{foo: bar }
%tag{ foo: bar }
With default space
style option: require a single space inside
hash attributes braces
%tag{ foo: bar }
With no_space
style option: require no space inside
hash attributes braces
%tag{foo: bar}
This offers the ability to ensure consistency of Haml hash attributes style with ruby hash literal style (compare with the Style/SpaceInsideHashLiteralBraces cop in Rubocop).
Tag names should not contain uppercase letters.
Bad
%BODY
Good
%body
This is a de facto standard in writing HAML documents as well as HTML in general, as it is easier to type and matches the convention of many developer tools. If you are writing HAML to output XML documents, however, it is a strict requirement.
HAML documents should not contain trailing whitespace (spaces or tabs) on any lines.
Avoid using unnecessary interpolation for inline tag content.
Bad
%tag #{expression}
Good: more concise
%tag= expression
Avoid outputting string expressions in Ruby when static text will suffice.
Bad
%tag= "Some #{interpolated} string"
Good: more concise
%tag Some #{interpolated} string
HAML gracefully handles string interpolation in static text, so you don't need to work with Ruby strings in order to use interpolation.