github.com/markets/invisible_captcha
Complete and flexible spam protection solution for Rails applications.
Invisible Captcha provides different techniques to protect your application against spambots.
The main protection is a solution based on the honeypot principle, which provides a better user experience since there are no extra steps for real users, only for the bots.
Essentially, the strategy consists on adding an input field :honey_pot: into the form that:
It also comes with:
Invisible Captcha is tested against Rails >= 5.2 and Ruby >= 2.7.
Add this line to your Gemfile and then execute bundle install:
gem 'invisible_captcha'
View code:
<%= form_for(@topic) do |f| %>
<%= f.invisible_captcha :subtitle %>
<!-- or -->
<%= invisible_captcha :subtitle, :topic %>
<% end %>
Controller code:
class TopicsController < ApplicationController
invisible_captcha only: [:create, :update], honeypot: :subtitle
end
This method will act as a before_action that triggers when spam is detected (honeypot field has some value). By default, it responds with no content (only headers: head(200)). This is a good default, since the bot will surely read the response code and will think that it has achieved to submit the form properly. But, anyway, you can define your own callback by passing a method to the on_spam option:
class TopicsController < ApplicationController
invisible_captcha only: [:create, :update], on_spam: :your_spam_callback_method
private
def your_spam_callback_method
redirect_to root_path
end
end
You should not name your method on_spam, as this will collide with an internal method of the same name.
Note that it is not mandatory to specify a honeypot attribute (neither in the view nor in the controller). In this case, the engine will take a random field from InvisibleCaptcha.honeypots. So, if you're integrating it following this path, in your form:
<%= form_tag(new_contact_path) do |f| %>
<%= invisible_captcha %>
<% end %>
In your controller:
invisible_captcha only: [:new_contact]
invisible_captcha sends all messages to flash[:error]. For messages to appear on your pages, add <%= flash[:error] %> to app/views/layouts/application.html.erb (somewhere near the top of your <body> element):
<!DOCTYPE html>
<html>
<head>
<title>Yet another Rails app</title>
<%= stylesheet_link_tag "application", media: "all" %>
<%= javascript_include_tag "application" %>
<%= csrf_meta_tags %>
</head>
<body>
<%= flash[:error] %>
<%= yield %>
</body>
</html>
You can place <%= flash[:error] %> next to :alert and :notice message types, if you have them in your app/views/layouts/application.html.erb.
NOTE: This gem relies on data set by the backend, so in order to properly work, your forms should be rendered by Rails. Forms generated via JavaScript are not going to work well.
This section contains a description of all plugin options and customizations.
You can customize:
sentence_for_humans: text for real users if input field was visible. By default, it uses I18n (see below).honeypots: collection of default honeypots. Used by the view helper, called with no args, to generate a random honeypot field name. By default, a random collection is already generated. As the random collection is stored in memory, it will not work if you are running multiple Rails instances behind a load balancer (see Multiple Rails instances). Beware that Chrome may ignore autocomplete="off". Thus, consider not to use field names, which would be autocompleted, like for example name, country.visual_honeypots: make honeypots visible, also useful to test/debug your implementation.timestamp_threshold: fastest time (in seconds) to expect a human to submit the form (see original article by Yoav Aner outlining the idea). By default, 4 seconds. NOTE: It's recommended to deactivate the autocomplete feature to avoid false positives (autocomplete="off").timestamp_enabled: option to disable the time threshold check at application level. Could be useful, for example, on some testing scenarios. By default, true.timestamp_error_message: flash error message thrown when form submitted quicker than the timestamp_threshold value. It uses I18n by default.injectable_styles: if enabled, you should call anywhere in your layout the following helper <%= invisible_captcha_styles %>. This allows you to inject styles, for example, in <head>. False by default, styles are injected inline with the honeypot.spinner_enabled: option to disable the IP spinner validation. By default, true.secret: customize the secret key to encode some internal values. By default, it reads the environment variable ENV['INVISIBLE_CAPTCHA_SECRET'] and fallbacks to random value. Be careful, if you are running multiple Rails instances behind a load balancer, use always the same value via the environment variable.To change these defaults, add the following to an initializer (recommended config/initializers/invisible_captcha.rb):
InvisibleCaptcha.setup do |config|
# config.honeypots << ['more', 'fake', 'attribute', 'names']
# config.visual_honeypots = false
# config.timestamp_threshold = 2
# config.timestamp_enabled = true
# config.injectable_styles = false
# config.spinner_enabled = true
# Leave these unset if you want to use I18n (see below)
# config.sentence_for_humans = 'If you are a human, ignore this field'
# config.timestamp_error_message = 'Sorry, that was too quick! Please resubmit.'
end
If you have multiple Rails instances running behind a load balancer, you have to share the same honeypots collection between the instances.
Either use a fixed collection or share them between the instances using Rails.cache:
InvisibleCaptcha.setup do |config|
config.honeypots = Rails.cache.fetch('invisible_captcha_honeypots') do
(1..20).map { InvisibleCaptcha.generate_random_honeypot }
end
end
Be careful also with the secret setting. Since it will be stored in-memory, if you are running this setup, the best idea is to provide the environment variable (ENV['INVISIBLE_CAPTCHA_SECRET']) from your infrastructure.
The invisible_captcha method accepts some options:
only: apply to given controller actions.except: exclude to given controller actions.honeypot: name of custom honeypot.scope: name of scope, ie: 'topic[subtitle]' -> 'topic' is the scope. By default, it's inferred from the controller_name.on_spam: custom callback to be called on spam detection.timestamp_enabled: enable/disable this technique at action level.on_timestamp_spam: custom callback to be called when form submitted too quickly. The default action redirects to :back printing a warning in flash[:error].timestamp_threshold: custom threshold per controller/action. Overrides the global value for InvisibleCaptcha.timestamp_threshold.prepend: the spam detection will run in a prepend_before_action if prepend: true, otherwise will run in a before_action.Using the view/form helper you can override some defaults for the given instance. Actually, it allows to change:
sentence_for_humans<%= form_for(@topic) do |f| %>
<%= f.invisible_captcha :subtitle, sentence_for_humans: "hey! leave this input empty!" %>
<% end %>
visual_honeypots<%= form_for(@topic) do |f| %>
<%= f.invisible_captcha :subtitle, visual_honeypots: true %>
<% end %>
You can also pass html options to the input:
<%= invisible_captcha :subtitle, :topic, id: "your_id", class: "your_class" %>
In addition to the on_spam controller callback, you can use the Active Support Instrumentation API to set up a global event handler that fires whenever spam is detected. This is useful for advanced logging, background processing, etc.
To set up a global event handler, subscribe to the invisible_captcha.spam_detected event in an initializer:
# config/initializers/invisible_captcha.rb
ActiveSupport::Notifications.subscribe('invisible_captcha.spam_detected') do |*args, data|
AwesomeLogger.warn(data[:message], data) # Log to an external logging service.
SpamRequest.create(data) # Record the blocked request in your database.
end
The data passed to the subscriber is hash containing information about the request that was detected as spam. For example:
{
message: "Honeypot param 'subtitle' was present.",
remote_ip: '127.0.0.1',
user_agent: 'Chrome 77',
controller: 'users',
action: 'create',
url: 'http://example.com/users',
params: {
topic: { subtitle: 'foo' },
controller: 'users',
action: 'create'
}
}
NOTE: params will be filtered according to your Rails.application.config.filter_parameters configuration, making them (probably) safe for logging. But always double-check that you're not inadvertently logging sensitive form data, like passwords and credit cards.
If you're using a Content Security Policy (CSP) in your Rails app, you will need to generate a nonce on the server, and pass nonce: true attribute to the view helper. Uncomment the following lines in your config/initializers/content_security_policy.rb file:
# Be sure to restart your server when you modify this file.
# If you are using UJS then enable automatic nonce generation
Rails.application.config.content_security_policy_nonce_generator = -> request { SecureRandom.base64(16) }
# Set the nonce only to specific directives
Rails.application.config.content_security_policy_nonce_directives = %w(style-src)
Note that if you are already generating nonce for scripts, you'd have to include script-src to content_security_policy_nonce_directives as well:
Rails.application.config.content_security_policy_nonce_directives = %w(script-src style-src)
And in your view helper, you need to pass nonce: true to the invisible_captcha helper:
<%= invisible_captcha nonce: true %>
NOTE: Content Security Policy can break your site! If you already run a website with third-party scripts, styles, images, and fonts, it is highly recommended to enable CSP in report-only mode and observe warnings as they appear. Learn more at MDN:
invisible_captcha tries to use I18n when it's available by default. The keys it looks for are the following:
en:
invisible_captcha:
sentence_for_humans: "If you are human, ignore this field"
timestamp_error_message: "Sorry, that was too quick! Please resubmit."
You can override the English ones in your i18n config files as well as add new ones for other locales.
If you intend to use I18n with invisible_captcha, you must not set sentence_for_humans or timestamp_error_message to strings in the setup phase.
If you're encountering unexpected behaviour while testing controllers that use the invisible_captcha action filter, you may want to disable timestamp check for the test environment. Add the following snippet to the config/initializers/invisible_captcha.rb file:
# Be sure to restart your server when you modify this file.
InvisibleCaptcha.setup do |config|
config.timestamp_enabled = !Rails.env.test?
end
Another option is to wait for the timestamp check to be valid:
# Maybe inside a before block
InvisibleCaptcha.init!
InvisibleCaptcha.timestamp_threshold = 1
# Before testing your controller action
sleep InvisibleCaptcha.timestamp_threshold
If you're using the "random honeypot" approach, you may want to set a known honeypot:
config.honeypots = ['my_honeypot_field'] if Rails.env.test?
And for the "spinner validation" check, you may want to disable it:
config.spinner_enabled = !Rails.env.test?
Or alternativelly, you should send a valid spinner value along your requests:
# RSpec example
session[:invisible_captcha_spinner] = '32ab649161f9f6faeeb323746de1a25d'
post :create, params: { topic: { title: 'foo' }, spinner: '32ab649161f9f6faeeb323746de1a25d' }
Any kind of idea, feedback or bug report are welcome! Open an issue or send a pull request.
Clone/fork this repository, start to hack on it and send a pull request.
Run the test suite:
$ bundle exec rspec
Run the test suite against all supported versions:
$ bundle exec appraisal install
$ bundle exec appraisal rspec
Run specs against specific version:
$ bundle exec appraisal rails-6.0 rspec
Start a sample Rails app (source code) with InvisibleCaptcha integrated:
$ bundle exec rake web # PORT=4000 (default: 3000)
Copyright (c) Marc Anguera. Invisible Captcha is released under the MIT License.
FAQs
Unknown package
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Socket's package search now displays weekly downloads for npm packages, helping developers quickly assess popularity and make more informed decisions.
Security News
A Stanford study reveals 9.5% of engineers contribute almost nothing, costing tech $90B annually, with remote work fueling the rise of "ghost engineers."
Research
Security News
Socket’s threat research team has detected six malicious npm packages typosquatting popular libraries to insert SSH backdoors.