Home

Awesome

HTMLProofer

If you generate HTML files, then this tool might be for you!

Project scope

HTMLProofer is a set of tests to validate your HTML output. These tests check if your image references are legitimate, if they have alt tags, if your internal links are working, and so on. It's intended to be an all-in-one checker for your output.

In scope for this project is any well-known and widely-used test for HTML document quality. A major use for this project is continuous integration -- so we must have reliable results. We usually balance correctness over performance. And, if necessary, we should be able to trace this program's detection of HTML errors back to documented best practices or standards, such as W3 specifications.

Third-party modules. We want this product to be useful for continuous integration so we prefer to avoid subjective tests which are prone to false positive results, such as spell checkers, indentation checkers, etc. If you want to work on these items, please see the section on custom tests and consider adding an implementation as a third-party module.

Advanced configuration. Most front-end developers can test their HTML using our command line program. Advanced configuration will require using Ruby.

Installation

Add this line to your application's Gemfile:

gem 'html-proofer'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install html-proofer

NOTE: When installation speed matters, set NOKOGIRI_USE_SYSTEM_LIBRARIES to true in your environment. This is useful for increasing the speed of your Continuous Integration builds.

What's tested?

Below is a mostly comprehensive list of checks that HTMLProofer can perform.

Images

img elements:

Links

a, link elements:

Scripts

script elements:

Favicon

OpenGraph

Usage

You can configure HTMLProofer to run on:

It can also run through the command-line.

Using in a script

  1. Require the gem.
  2. Generate some HTML.
  3. Create a new instance of the HTMLProofer on your output folder.
  4. Call proofer.run on that path.

Here's an example:

require "html-proofer"
require "html/pipeline"
require "find"

# make an out dir
Dir.mkdir("out") unless File.exist?("out")

pipeline = HTML::Pipeline.new([
  HTML::Pipeline::MarkdownFilter,
  HTML::Pipeline::TableOfContentsFilter,
],
gfm: true)

# iterate over files, and generate HTML from Markdown
Find.find("./docs") do |path|
  next unless File.extname(path) == ".md"
  contents = File.read(path)
  result = pipeline.call(contents)

  File.open("out/#{path.split("/").pop.sub(".md", ".html")}", "w") { |file| file.write(result[:output].to_s) }
end

# test your out dir!
HTMLProofer.check_directory("./out").run

Checking a single file

If you simply want to check a single file, use the check_file method:

HTMLProofer.check_file("/path/to/a/file.html").run

Checking directories

If you want to check a directory, use check_directory:

HTMLProofer.check_directory("./out").run

If you want to check multiple directories, use check_directories:

HTMLProofer.check_directories(["./one", "./two"]).run

Checking an array of links

With check_links, you can also pass in an array of links:

HTMLProofer.check_links(["https://github.com", "https://jekyllrb.com"]).run

Swapping information

Sometimes, the information in your HTML is not the same as how your server serves content. In these cases, you can use swap_urls to map the URL in a file to the URL you'd like it to become. For example:

run_proofer(file, :file, swap_urls: { %r{^https//placeholder.com} => "https://website.com" })

In this case, any link that matches the ^https://placeholder.com will be converted to https://website.com.

A similar swapping process can be done for attributes:

run_proofer(file, :file, swap_attributes: { "img"  => [["data-src", "src"]] })

In this case, we are telling HTMLProofer that, for any img tag detected, for any src attribute, pretend it's actually the src attribute instead. Since the value is an array of arrays, you can pass in as many attribute swaps as you need for each element.

Using on the command-line

You'll also get a new program called htmlproofer with this gem. Terrific!

Pass in options through the command-line as flags, like this:

htmlproofer --extensions .html.erb ./out

Use htmlproofer --help to see all command line options.

Special cases for the command-line

For options which require an array of input, surround the value with quotes, and don't use any spaces. For example, to exclude an array of HTTP status code, you might do:

htmlproofer --ignore-status-codes "999,401,404" ./out

For something like url-ignore, and other options that require an array of regular expressions, you can pass in a syntax like this:

htmlproofer --ignore-urls "/www.github.com/,/foo.com/" ./out

Since swap_urls is a bit special, you'll pass in a pair of RegEx:String values. The escape sequences \: should be used to produce literal :s htmlproofer will figure out what you mean.

htmlproofer --swap-urls "wow:cow,mow:doh" --extensions .html.erb --ignore-urls www.github.com ./out

Some configuration options, such as --typheous, --cache, or --swap-attributes, require well-formatted JSON.

Adjusting for a baseurl

If your Jekyll site has a baseurl configured, you'll need to adjust the generated url validation to cope with that. The easiest way is using the swap_urls option.

For a site.baseurl value of /BASEURL, here's what that looks like on the command line:

htmlproofer --assume-extension ./_site --swap-urls '^/BASEURL/:/'

or in your Rakefile

require "html-proofer"

task :test do
  sh "bundle exec jekyll build"
  options = { swap_urls: "^/BASEURL/:/" }
  HTMLProofer.check_directory("./_site", options).run
end

Using through Docker

If you have trouble with (or don't want to) install Ruby/Nokogumbo, the command-line tool can be run through Docker. See klakegg/html-proofer for more information.

Ignoring content

Add the data-proofer-ignore attribute to any tag to ignore it from every check.

<a href="https://notareallink" data-proofer-ignore>Not checked.</a>

This can also apply to parent elements, all the way up to the <html> tag:

<div data-proofer-ignore>
  <a href="https://notareallink">Not checked because of parent.</a>
</div>

Ignoring new files

Say you've got some new files in a pull request, and your tests are failing because links to those files are not live yet. One thing you can do is run a diff against your base branch and explicitly ignore the new files, like this:

directories = ['content']
merge_base = %x(git merge-base origin/production HEAD).chomp
diffable_files = %x(git diff -z --name-only --diff-filter=AC #{merge_base}).split("\0")
diffable_files = diffable_files.select do |filename|
  next true if directories.include?(File.dirname(filename))

  filename.end_with?(".md")
end.map { |f| Regexp.new(File.basename(f, File.extname(f))) }

HTMLProofer.check_directory("./output", { ignore_urls: diffable_files }).run

Configuration

The HTMLProofer constructor takes an optional hash of additional options:

OptionDescriptionDefault
allow_hash_hrefIf true, assumes href="#" anchors are validtrue
allow_missing_hrefIf true, does not flag a tags missing href. In HTML5, this is technically allowed, but could also be human error.false
assume_extensionAutomatically add specified extension to files for internal links, to allow extensionless URLs (as supported by most servers).html
checksAn array of Strings indicating which checks you want to run['Links', 'Images', 'Scripts']
check_external_hashChecks whether external hashes exist (even if the webpage exists)true
check_internal_hashChecks whether internal hashes exist (even if the webpage exists)true
check_sriCheck that <link> and <script> external resources use SRIfalse
directory_index_fileSets the file to look for when a link refers to a directory.index.html
disable_externalIf true, does not run the external link checkerfalse
enforce_httpsFails a link if it's not marked as https.true
extensionsAn array of Strings indicating the file extensions you would like to check (including the dot)['.html']
ignore_empty_altIf true, ignores images with empty/missing alt tags (in other words, <img alt> and <img alt=""> are valid; set this to false to flag those)true
ignore_filesAn array of Strings or RegExps containing file paths that are safe to ignore.[]
ignore_empty_mailtoIf true, allows mailto: hrefs which do not contain an email address.false
ignore_missing_altIf true, ignores images with missing alt tagsfalse
ignore_status_codesAn array of numbers representing status codes to ignore.[]
ignore_urlsAn array of Strings or RegExps containing URLs that are safe to ignore. This affects all HTML attributes, such as alt tags on images.[]
log_levelSets the logging level, as determined by Yell. One of :debug, :info, :warn, :error, or :fatal.:info
only_4xxOnly reports errors for links that fall within the 4xx status code range.false
root_dirThe absolute path to the directory serving your html-files.""
swap_attributesJSON-formatted config that maps element names to the preferred attribute to check{}
swap_urlsA hash containing key-value pairs of RegExp => String. It transforms URLs that match RegExp into String via gsub.{}

In addition, there are a few "namespaced" options. These are:

Configuring Typhoeus and Hydra

Typhoeus is used to make fast, parallel requests to external URLs. You can pass in any of Typhoeus' options for the external link checks with the options namespace of :typhoeus. For example:

HTMLProofer.new("out/", { extensions: [".htm"], typhoeus: { verbose: true, ssl_verifyhost: 2 } })

This sets HTMLProofer's extensions to use .htm, gives Typhoeus a configuration for it to be verbose, and use specific SSL settings. Check the Typhoeus documentation for more information on what options it can receive.

You can similarly pass in a :hydra option with a hash configuration for Hydra.

The default value is:

{
  typhoeus:
  {
    followlocation: true,
    connecttimeout: 10,
    timeout: 30,
  },
  hydra: { max_concurrency: 50 },
}

On the CLI, you can provide the --typhoeus or hydra arguments to set the configurations. This is parsed using JSON.parse and mapped on top of the default configuration values so that they can be overridden.

Setting before-request callback

You can provide a block to set some logic before an external link is checked. For example, say you want to provide an authentication token every time a GitHub URL is checked. You can do that like this:

proofer = HTMLProofer.check_directory(item, opts)
proofer.before_request do |request|
  request.options[:headers]["Authorization"] = "Bearer <TOKEN>" if request.base_url == "https://github.com"
end
proofer.run

The Authorization header is being set if and only if the base_url is https://github.com, and it is excluded for all other URLs.

Configuring caching

Checking external URLs can slow your tests down. If you'd like to speed that up, you can enable caching for your external and internal links. Caching simply means to skip link checking for links that are valid for a certain period of time.

You can enable caching for this by passing in the configuration option :cache, with a hash containing a single key, :timeframe. :timeframe defines the length of time the cache will be used before the link is checked again. The format of :timeframe is a hash containing two keys, external and internal. Each of these contains a number followed by a letter indicating the length of time:

For example, passing the following options means "recheck external links older than thirty days":

{ cache: { timeframe: { external: "30d" } } }

And the following options means "recheck internal links older than two weeks":

{ cache: { timeframe: { internal: "2w" } } }

Naturally, to support both internal and external link caching, both keys would need to be provided. The following checks external links every two weeks, but internal links only once a week:

{ cache: { timeframe: { external: "2w", internal: "1w" } } }

You can change the filename or the directory where the cache file is kept by also providing the storage_dir key:

{ cache: { cache_file: "stay_cachey.json", storage_dir: "/tmp/html-proofer-cache-money" } }

Links that were failures are kept in the cache and always rechecked. If they pass, the cache is updated to note the new timestamp.

The cache operates on external links only.

If caching is enabled, HTMLProofer writes to a log file called tmp/.htmlproofer/cache.log. You should probably ignore this folder in your version control system.

On the CLI, you can provide the --cache argument to set the configuration. This is parsed using JSON.parse and mapped on top of the default configuration values so that they can be overridden.

Caching with continuous integration

Enable caching in your continuous integration process. It will make your builds faster.

In GitHub Actions:

Add this step to your build workflow before HTMLProofer is run:

- name: Cache HTMLProofer
  id: cache-htmlproofer
  uses: actions/cache@v2
  with:
    path: tmp/.htmlproofer
    key: ${{ runner.os }}-htmlproofer

Also make sure that your later step which runs HTMLProofer will not return a failed shell status. You can try something like html-proof ... || true. Because a failed step in GitHub Actions will skip all later steps.

In Travis:

If you want to enable caching with Travis CI, be sure to add these lines into your .travis.yml file:

cache:
  directories:
    - $TRAVIS_BUILD_DIR/tmp/.htmlproofer

For more information on using HTML-Proofer with Travis CI, see this wiki page.

Logging

HTML-Proofer can be as noisy or as quiet as you'd like. If you set the :log_level option, you can better define the level of logging.

Custom tests

Want to write your own test? Sure, that's possible!

Just create a class that inherits from HTMLProofer::Check. This subclass must define one method called run. This is called on your content, and is responsible for performing the validation on whatever elements you like. When you catch a broken issue, call add_failure(message, line: line, content: content) to explain the error. line refers to the line numbers, and content is the node content of the broken element.

If you're working with the element's attributes (as most checks do), you'll also want to call create_element(node) as part of your suite. This constructs an object that contains all the attributes of the HTML element you're iterating on, and can also be used directly to call add_failure(message, element: element).

Here's an example custom test demonstrating these concepts. It reports mailto links that point to octocat@github.com:

class MailToOctocat < HTMLProofer::Check
  def mailto_octocat?
    @link.url.raw_attribute == "mailto:octocat@github.com"
  end

  def run
    @html.css("a").each do |node|
      @link = create_element(node)

      next if @link.ignore?

      return add_failure("Don't email the Octocat directly!", element: @link) if mailto_octocat?
    end
  end
end

Don't forget to include this new check in HTMLProofer's options, for example:

# removes default checks and just runs this one
HTMLProofer.check_directories(["out/"], { checks: ["MailToOctocat"] })

See our list of third-party custom classes and add your own to this list.

Reporting

By default, HTML-Proofer has its own reporting mechanism to print errors at the end of the run. You can choose to use your own reporter by passing in your own subclass of HTMLProofer::Reporter:

proofer = HTMLProofer.check_directory(item, opts)
proofer.reporter = MyCustomReporter.new(logger: proofer.logger)
proofer.run

Your custom reporter must implement the report function which implements the behavior you wish to see. The logger kwarg is optional.

Troubleshooting

Here are some brief snippets identifying some common problems that you can work around. For more information, check out our wiki.

Our wiki page on using HTML-Proofer with Travis CI might also be useful.

Ignoring SSL certificates

To ignore SSL certificates, turn off Typhoeus' SSL verification:

HTMLProofer.check_directory("out/", {
  typhoeus: {
    ssl_verifypeer: false,
    ssl_verifyhost: 0,
},
}).run

User-Agent

To change the User-Agent used by Typhoeus:

HTMLProofer.check_directory("out/", {
  typhoeus: {
    headers: { "User-Agent" => "Mozilla/5.0 (compatible; My New User-Agent)" },
  }
}).run

Alternatively, you can specify these options on the command-line with:

htmlproofer --typhoeus='{"headers":{"User-Agent":"Mozilla/5.0 (compatible; My New User-Agent)"}}'

Cookies

Sometimes links fail because they don't have access to cookies. To fix this you can create a .cookies file using the following snippets:

HTMLProofer.check_directory("out/", {
  typhoeus: {
    cookiefile: ".cookies",
    cookiejar: ".cookies",
  }
}).run
htmlproofer --typhoeus='{"cookiefile":".cookies","cookiejar":".cookies"}'

Regular expressions

To exclude urls using regular expressions, include them between forward slashes and don't quote them:

HTMLProofer.check_directories(["out/"], {
  ignore_urls: [/example.com/],
}).run

Real-life examples

ProjectRepositoryNotes
Jekyll's websitejekyll/jekyllA separate script calls htmlproofer and this used to be called from Circle CI
Raspberry Pi's documentationraspberrypi/documentation
Squeak's websitesqueak-smalltalk/squeak.org
Atom Flight Manualatom/flight-manual.atom.io
HTML Website Templatefulldecent/html-website-templateA starting point for websites, uses a Rakefile and Travis configuration to call preconfigured testing
Project Calico Documentationprojectcalico/calicoSimple integration with Jekyll and Docker using a Makefile
GitHub does dotfilesdotfiles/dotfiles.github.comUses the proof-html GitHub action