ERB Lint

erb_lint is a tool to help lint your ERB or HTML files using the included linters or by writing your own.

• Ruby 2.3.0+

• This is due to use of the safe navigation operator (&.)
• This is also due to the use of the tilde-heredoc <<~ syntax in some tests.

gem install erb_lint

...or add the following to your Gemfile and run bundle install:

gem 'erb_lint', require: false

Create a .erb_lint.yml file in your project, with the following structure:

---
EnableDefaultLinters: true
linters:
  ErbSafety:
    enabled: true
    better_html_config: .better-html.yml
  Rubocop:
    enabled: true
    rubocop_config:
      inherit_from:
        - .rubocop.yml

See below for linter-specific configuration options.

This gem provides a command-line interface which can be run like so:

1. Run erb_lint [options] if the gem is installed standalone.
2. Run bundle exec erb_lint [options] if the gem is installed as a Gemfile dependency for your app.

For example, erb_lint --lint-all --enable-all-linters will run all available linters on all ERB files in the current directory or its descendants (**/*.html{+*,}.erb).

If you want to change the glob & exclude that is used, you can configure it by adding it to your config file as follows:

---
glob: "**/*.{html,text,js}{+*,}.erb"
exclude:
  - '**/vendor/**/*'
  - '**/node_modules/**/*'
linters:
  ErbSafety:
    enabled: true
    better_html_config: .better-html.yml
  Rubocop:
    enabled: true
    rubocop_config:
      inherit_from:
        - .rubocop.yml

Make sure to add **/ to exclude patterns; it matches the target files' absolute paths.

EnableDefaultLinters: enables or disables default linters. Default linters are enabled by default.

You can disable a rule by placing a disable comment in the following format:

Comment on offending lines

<hr /> <%# erb_lint:disable SelfClosingTag %>

To raise an error when there is a useless disable comment, enable NoUnusedDisable.

To disable inline comments and report all offenses, set --disable-inline-configs option.

You can specify the exclude patterns both of global and lint-local.

---
exclude:
  - '**/global-lib/**/*'
linters:
  ErbSafety:
    exclude:
      - '**/local-lib/**/*'

DeprecatedClasses will find all classes used on HTML elements and report any classes that violate the rule set that you provide.

A rule_set is specified as a list, each with a set of deprecated classes and a corresponding suggestion to use as an alternative.

Example configuration:

---
linters:
  DeprecatedClasses:
    enabled: true
    exclude:
      - 'app/views/shared/deprecated/**'
    addendum: "See UX wiki for help."
    rule_set:
      - deprecated: ['badge[-_\w]*']
        suggestion: "Use the ui_badge() component instead."

You can specify an addendum to be added to the end of each violation. The error message format is: "Deprecated class ... #{suggestion}" or "Deprecated class ... #{suggestion} #{addendum}" if an addendum is present.

Files must have a final newline. This results in better diffs when adding lines to the file, since SCM systems such as git won't think that you touched the last line.

You can customize whether or not a final newline exists with the present option.

Example configuration:

---
linters:
  FinalNewline:
    enabled: true

Runs the checks provided by better-html's erb safety test helper.

When using ERB interpolations in javascript contexts, this linter enforces the usage of safe helpers such as .to_json. See better-html's readme for more information.

Any ERB statement that does not call a safe helper is deemed unsafe and a violation is shown.

For example:

Not allowed ❌
<a onclick="alert(<%= some_data %>)">

Allowed ✅
<a onclick="alert(<%= some_data.to_json %>)">

Not allowed ❌
<script>var myData = <%= some_data %>;</script>

Allowed ✅
<script>var myData = <%= some_data.to_json %>;</script>

Example configuration:

---
linters:
  ErbSafety:
    enabled: true
    better_html_config: .better-html.yml

Runs RuboCop on all ruby statements found in ERB templates. The RuboCop configuration that erb_lint uses can inherit from the configuration that the rest of your application uses. erb_lint can be configured independently however, as it will often be necessary to disable specific RuboCop rules that do not apply to ERB files.

Note: Each ruby statement (between ERB tags <% ... %>) is parsed and analyzed independently of each other. Any rule that requires a broader context can trigger false positives (e.g. Lint/UselessAssignment will complaint for an assignment even if used in a subsequent ERB tag).

Example configuration:

---
linters:
  Rubocop:
    enabled: true
    rubocop_config:
      inherit_from:
        - .rubocop.yml
      Layout/InitialIndentation:
        Enabled: false
      Layout/LineLength:
        Enabled: false
      Layout/TrailingEmptyLines:
        Enabled: false
      Layout/TrailingWhitespace:
        Enabled: false
      Naming/FileName:
        Enabled: false
      Style/FrozenStringLiteralComment:
        Enabled: false
      Lint/UselessAssignment:
        Enabled: false
      Rails/OutputSafety:
        Enabled: false

The cops disabled in the example configuration above provide a good starting point.

This linter prevents the usage of certain types of HTML <input> without an autocomplete argument: color, date, datetime-local, email, month, number, password, range, search, tel, text, time, url, or week. The HTML autocomplete helps users to complete filling in forms by using data stored in the browser. This is particularly useful for people with motor disabilities or cognitive impairment who may have difficulties filling out forms online.

Bad ❌
<input type="email" ...>
Good ✅
<input type="email" autocomplete="nope" ...>
<input type="email" autocomplete="email" ...>

Trimming at the right of an ERB tag can be done with either =%> or -%>, this linter enforces one of these two styles.

Example configuration:

---
linters:
  RightTrim:
    enabled: true
    enforced_style: '-'

Enforce a single space after <% and before %> in the ERB source. This linter ignores opening ERB tags (<%) that are followed by a newline, and closing ERB tags (%>) that are preceded by a newline.

Bad ❌
<%foo%>
<%=foo-%>

Good ✅
<% foo %>

<%
  foo
%>

Example configuration:

---
linters:
  SpaceAroundErbTag:
    enabled: true

This linter prevents the usage of Rails' javascript_tag helper in ERB templates.

The html parser used in this gem knows not to look for html tags within certain other tags like script, style, and others. The html parser does this to avoid confusing javascript expressions like if (1<a || b>1) for a malformed html tag. Using the javascript_tag in a ERB template prevents the parser from recognizing the change of parsing context and may fail or produce erroneous output as a result.

Bad ❌
<%= javascript_tag(content, defer: true) %>
Good ✅
<script defer="true"><%== content %></script>

Bad ❌
<%= javascript_tag do %>
  alert(1)
<% end %>
Good ✅
<script>
  alert(1)
</script>

The autocorrection rule adds //<![CDATA[ and //]]> markers to the existing script, as this is the default behavior for javascript_tag. This can be disabled by changing the correction_style linter option from cdata to plain.

Example configuration:

---
linters:
  NoJavascriptTagHelper:
    enabled: true
    correction_style: 'plain'

This linter prevents the usage of HTML <script>, Rails javascript_tag, javascript_include_tag and javascript_pack_tag without a nonce argument. The purpose of such a check is to ensure that when content securty policy is implemented in an application, there is a means of discovering tags that need to be updated with a nonce argument to enable script execution at application runtime.

Bad ❌
<script>
    alert(1)
</script>
Good ✅
<script nonce="<%= request.content_security_policy_nonce %>" >
    alert(1)
</script>

Bad ❌
<%= javascript_tag do -%>
  alert(1)
<% end -%>
Good ✅
<%= javascript_tag nonce: true do -%>
  alert(1)
<% end -%>

Bad ❌
<%= javascript_include_tag "script" %>
Good ✅
<%= javascript_include_tag "script", nonce: true %>

Bad ❌
<%= javascript_pack_tag "script" %>
Good ✅
<%= javascript_pack_tag "script", nonce: true %>

This linter enforces self closing tag styles for void elements.

The void elements are area, base, br, col, embed, hr, img, input, keygen, link, menuitem, meta, param, source, track, and wbr.

If enforced_style is set to always (XHTML style):

Bad ❌
<link rel="stylesheet" type="text/css" href="styles.css">
Good ✅
<img src="someimage.png" alt="Some Image" />

If enforced_style is set to never (HTML5 style):

Bad ❌
<hr />
Good ✅
<meta charset="UTF-8">

Example configuration:

---
linters:
  SelfClosingTag:
    enabled: true
    enforced_style: 'always'

This linter prevent the addition of <script> tags that have type attributes that are not in a white-list of allowed values.

It is common practice for web developers to use <script> tags with non-executable type attributes, such as application/json or text/html to pass arbitrary data into an html page. Despite not being executable, these tags are subject to the same parsing quirks as executable script tags, and it is therefore more difficult to prevent security issues from creeping in. Consider for instance an application where it is possible to inject the string </script><script> unescaped into a text/html tag, the application would be vulnerable to XSS.

This pattern can easily be replaced by <div> tags with data attributes that can just as easily be read from javascript, and have the added benefit of being safer. When content_tag(:div) or tag.div() is used to pass arbitrary user data into the html document, it becomes much harder to inadvertently introduce a security issue.

It may also be desirable to avoid typos in type attributes.

Bad ❌
<script type="text/javacsrïpt"></script>
Good ✅
<script type="text/javascript"></script>

By default, this linter allows the type attribute to be omitted, as the behavior in browsers is to consider <script> to be the same as <script type="text/javascript">. When the linter is configured with allow_blank: false, instances of <script> tags without a type will be auto-corrected to <script type="text/javascript">.

It may also be desirable to disallow <script> tags from appearing anywhere in your application. For instance, Rails applications can benefit from serving static javascript code from the asset pipeline, as well as other security benefits. The disallow_inline_scripts: true config option may be used for that purpose.

Example configuration:

---
linters:
  AllowedScriptType:
    enabled: true
    allowed_types:
      - 'application/json'
      - 'text/javascript'
      - 'text/html'
    allow_blank: false
    disallow_inline_scripts: false

HardCodedString warns if there is a visible hardcoded string in the DOM. It does not check for a hardcoded string nested inside a JavaScript tag.

Example configuration:

---
linters:
  HardCodedString:
    enabled: true
    corrector:
      path: path/to/corrector_file.rb
      name: I18nCorrector
      i18n_load_path: config/locales/en.yml

To leverage this linter's autocorrect feature, you must define the configuration for its corrector, such that its path, class name, and i18n_load_path are resolvable.

Below is an example corrector file. For your project, the actual details of the autocorrect method are left up to how you want to correct those offenses.

class I18nCorrector
  attr_reader :node

  def initialize(node, filename, i18n_load_path, range)
  end

  def autocorrect(tag_start:, tag_end:)
    ->(corrector) do
      node
    end
  end
end

This linter enforces the use of the correct ERB comment syntax, since Ruby comments (<% # comment %> with a space) are not technically valid ERB comments.

Bad ❌
<% # This is a Ruby comment %>
Good ✅
<%# This is an ERB comment %>

Bad ❌
<% # This is a Ruby comment; it can fail to parse. %>
Good ✅
<%# This is an ERB comment; it is parsed correctly. %>

Good ✅
<%
  # This is a multi-line ERB comment.
%>

erb_lint allows you to create custom linters specific to your project. It will load linters from the .erb_linters directory in the root of your repository. See the linters directory for examples of how to write linters.

# .erb_linters/custom_linter.rb

module ERBLint
  module Linters
    class CustomLinter < Linter
      include LinterRegistry

      class ConfigSchema < LinterConfig
        property :custom_message, accepts: String
      end
      self.config_schema = ConfigSchema

      def run(processed_source)
        unless processed_source.file_content.include?('this file is fine')
          add_offense(
            processed_source.to_source_range(0 ... processed_source.file_content.size),
            "This file isn't fine. #{@config.custom_message}"
          )
        end
      end
    end
  end
end

By default, this linter would be disabled. You can enable it by adding an entry to .erb_lint.yml:

---
linters:
  CustomLinter:
    enabled: true
    custom_message: We suggest you change this file.

Test your linter by running erb_lint's command-line interface:

bundle exec erb_lint --enable-linters custom_linter --lint-all

Running this on a random project might yield this output:

Linting 15 files with 1 linters...

This file isn't fine. We suggest you change this file.
In file: app/views/layouts/application.html.erb:1

Errors were found in ERB files

To write a linter that can autocorrect offenses it detects, simply add an autocorrect method that returns a callable. The callable is called with an instance of RuboCop::Cop::Corrector as argument, and therefore erb_lint correctors work exactly as RuboCop correctors do.

def autocorrect(_processed_source, offense)
  lambda do |corrector|
    corrector.insert_after(offense.source_range, "this file is fine")
  end
end

You can change the output format of ERB Lint by specifying formatters with the -f/--format option.

$ erb_lint
Linting 8 files with 12 linters...

Remove multiple trailing newline at the end of the file.
In file: app/views/users/show.html.erb:95

Remove newline before `%>` to match start of tag.
In file: app/views/subscriptions/index.html.erb:38

2 error(s) were found in ERB files

erb_lint --format compact
Linting 8 files with 12 linters...
app/views/users/show.html.erb:95:0: Remove multiple trailing newline at the end of the file.
app/views/users/_graph.html.erb:27:37: Extra space detected where there should be no space
2 error(s) were found in ERB files

erb_lint --format junit
<?xml version="1.0" encoding="UTF-8"?>
<testsuite name="erb_lint" tests="2" failures="2">
  <properties>
    <property name="erb_lint_version" value="%{erb_lint_version}"/>
    <property name="ruby_engine" value="%{ruby_engine}"/>
    <property name="ruby_version" value="%{ruby_version}"/>
    <property name="ruby_patchlevel" value="%{ruby_patchlevel}"/>
    <property name="ruby_platform" value="%{ruby_platform}"/>
  </properties>
  <testcase name="app/views/subscriptions/_loader.html.erb" file="app/views/subscriptions/_loader.html.erb" lineno="1">
    <failure message="SpaceInHtmlTag: Extra space detected where there should be no space." type="SpaceInHtmlTag">
      <![CDATA[SpaceInHtmlTag: Extra space detected where there should be no space. at app/views/subscriptions/_loader.html.erb:1:7]]>
    </failure>
  </testcase>
  <testcase name="app/views/application/index.html.erb" file="app/views/subscriptions/_menu.html.erb"/>
</testsuite>

Used by GitLab Code Quality.

[
   {
      "description":"Extra space detected where there should be no space.",
      "check_name":"SpaceInHtmlTag",
      "fingerprint":"5a259c7cafa2c9ca229dfd7d21536698",
      "severity":"info",
      "location":{
         "path":"app/views/subscriptions/_loader.html.erb",
         "lines":{
            "begin":1,
            "end":1
         }
      }
   },
   {
      "description":"Remove newline before `%\u003e` to match start of tag.",
      "check_name":"ClosingErbTagIndent",
      "fingerprint":"60b4ed2120c7abeebebb43fba4a19559",
      "severity":"warning",
      "location":{
         "path":"app/views/subscriptions/_loader.html.erb",
         "lines":{
            "begin":52,
            "end":54
         }
      }
   }
]

The cache is currently opt-in - to turn it on, use the --cache option:

erb_lint --cache ./app
Cache mode is on
Linting 413 files with 15 linters...
File names pruned from the cache will be logged

No errors were found in ERB files

Cached lint results are stored in the .erb_lint_cache directory by default, though a custom directory can be provided via the --cache-dir option. Cache filenames are computed with a hash of information about the file and erb_lint settings. These files store instance attributes of the CachedOffense object, which only contain the Offense attributes necessary to restore the results of running erb_lint for output. The cache also automatically prunes outdated files each time it's run.

You can also use the --clear-cache option to delete the cache file directory.

This project is released under the MIT license.