How To

This page contains the practical tips and examples to get the job done with Pagy. If there is something missing, or some topic that you think should be added, fixed or explained better, please open an issue.

Quick Start

1. Install Pagy:

   • If you use Bundler, add the gem in the Gemfile, optionally avoiding the next mayor version with breaking changes (e.g. ‘~> 3.5’ see RubyGem Specifiers):

       gem 'pagy', '~> 3.5' # omit patch digit and use the latest if possible
     

   • If you don’t use Bundler, install and require the Pagy gem:

       $ gem install pagy
     

       require 'pagy'
     

2. Add the pagy.rb configuration file to your app:

   • If you use Rails, put it into the config/initializers dir

   • if you don’t use Rails it doesn’t matter where you put it, as long as you require it when your app starts

3. Include the backend in some controller (e.g. ApplicationController in Rails):

    include Pagy::Backend
   

4. Use the pagy method in some action:

    @pagy, @records = pagy(Product.some_scope)
   

5. Render the pagination:

   • Option A: if your app renders the views server-side:

     1. Include the frontend in some helper (e.g. ApplicationHelper in Rails):

         include Pagy::Frontend
        

     2. Render the navigation links in some view:

        • with a fast helper (also styled for bootstrap, bulma, foundation, materialize, semantic, uikit and available in different flavors (static, responsive, compact, etc.)

            <%== pagy_nav(@pagy) %>
          

        • or with an easy customizable template:

            <%== render partial: 'pagy/nav', locals: {pagy: @pagy} %>
          

   • Option B: if your app renders the views with a javascript framework (e.g. Vue.js, react.js, …), you don’t need the include Pagy::Frontend in ApplicationHelper, instead:

     1. require the metadata extra by uncommenting the following line in your pagy.rb file:

         require 'pagy/extras/metadata'
        

     2. add the metadata to your JSON response:

        render json: { data: @records,
                       pagy: pagy_metadata(@pagy) }
        

   • Option C: if your app is an API service consumed by some client and doesn’t provide any UI on its own, you don’t need the include Pagy::Frontend in ApplicationHelper, instead:

     1. require the headers extra by uncommenting it in your pagy.rb file:

         require 'pagy/extras/headers'
        

     2. add the pagination headers to your responses:

         after_action { pagy_headers_merge(@pagy) if @pagy }
        

     3. render your JSON response as usual:

         render json: { data: @records }
        

Global Configuration

Unlike the other pagination gems, Pagy is very modular so it doesn’t load nor execute unnecessary code in your app. Every feature that is not strictly needed for the basic pagination can be explicitly required in your initializer file.

Basic pagination should work out of the box for most Rack based apps (e.g. Rails) even without configuring/requiring anything, however you can tweak all its features and all the extras by loading a pagy.rb initializer file.

You can copy the comprehensive and annotated pagy.rb initializer and uncomment and edit what you may need. The file contains also all the relevant documentation links.

Environment Assumptions

• ruby 1.9+ or jruby 1.7+

Out of the box assumptions

Pagy works out of the box assuming that:

• You are using a Rack based framework
• The collection to paginate is an ORM collection (e.g. ActiveRecord scope)
• The controller where you include Pagy::Backend responds to a params method
• The view where you include Pagy::Frontend responds to a request method returning a Rack::Request instance.

Any other scenario assumptions

Pagy can also work in any other scenario assuming that:

• If your framework doesn’t have a params method you may need to define the params method or override the pagy_get_vars (which uses the params method) in your controller
• If the collection you are paginating doesn’t respond to offset and limit you may need to override the pagy_get_items method in your controller (to get the items out of your specific collection) or use a specific extra if available (e.g. array, searchkick, elasticsearch_rails)
• If your framework doesn’t have a request method you may need to override the pagy_url_for (which uses Rack and request) in your view

Notice: the total overriding you may need is usually just a handful of lines at worse, and it doesn’t need monkey patching or writing any sub-class or module.

Items per page

You can control the items per page with the items variable. (Default 20)

You can set its default in the pagy.rb initializer (see Configuration). For example:

Pagy::VARS[:items] = 25

You can also pass it as an instance variable to the Pagy.new method or to the pagy controller method:

@pagy, @records = pagy(Product.some_scope, items: 30)

If you want to allow the client to request a custom number of items per page - useful with APIs or highly user-customizable apps - take a look at the items extra. It manages the number of items per page requested with the params, and offers a ready to use selector UI.

Controlling the page links

You can control the number and position of the page links in the navigation through the :size variable. It is an array of 4 integers that specify which and how many page links to show.

The default is [1,4,4,1], which means that you will get 1 initial page, 4 pages before the current page, 4 pages after the current page, and 1 final page.

As usual you can set the :size variable as a global default by using the Pagy::VARS hash or pass it directly to the pagy method.

The navigation links will contain the number of pages set in the variables:

size[0]…size[1] current page size[2]…size[3]

For example:

pagy = Pagy.new count:1000, page: 10, size: [3,4,4,3] # etc
pagy.series
#=> [1, 2, 3, :gap, 6, 7, 8, 9, "10", 11, 12, 13, 14, :gap, 48, 49, 50]

As you can see by the result of the series method, you get 3 initial pages, 1 :gap (series interrupted), 4 pages before the current page, the current :page (which is a string), 4 pages after the current page, another :gap and 3 final pages.

You can easily try different options (also asymmetrical) in a console by changing the :size. Just check the series array to see what it contains when used in combination with different core variables.

Skipping the page links

If you want to skip the generation of the page links, just set the :size variable to an empty array:

For example:

pagy = Pagy.new count:1000, size: [] # etc
pagy.series
#=> []

Passing the page number

You don’t need to explicitly pass the page number to the pagy method, because it is pulled in by the pagy_get_vars (which is called internally by the pagy method). However you can force a page number by just passing it to the pagy method. For example:

@pagy, @records = pagy(my_scope, page: 3)

That will explicitly set the :page variable, overriding the default behavior (which usually pulls the page number from the params[:page]).

Customizing the page param

Pagy uses the :page_param variable to determine the param it should get the page number from and create the URL for. Its default is set as Pagy::VARS[:page_param] = :page, hence it will get the page number from the params[:page] and will create page URLs like ./?page=3 by default.

You may want to customize that, for example to make it more readable in your language, or because you need to paginate different collections in the same action. Depending on the scope of the customization, you have a couple of options:

1. Pagy::VARS[:page_param] = :custom_param will be used as the global default
2. pagy(scope, page_param: :custom_param) or Pagy.new(count:100, page_param: :custom_param) will be used for a single instance (overriding the global default)

You can also override the pagy_get_vars if you need some special way to get the page number.

Customizing the link attributes

If you need to customize some HTML attribute of the page links, you may not need to override the pagy_nav* helper. It might be enough to pass some extra attribute string with the :link_extra variable. For example:

# for all the Pagy instances
Pagy::VARS[:link_extra] = 'data-remote="true" class="my-class"'

# for a single Pagy instance (if you use the Pagy::Backend#pagy method)
@pagy, @records = pagy(my_scope, link_extra: 'data-remote="true" class="my-class"')

# or directly to the constructor
pagy = Pagy.new(count: 1000, link_extra: 'data-remote="true" class="my-class"')

IMPORTANT: For performance reasons, the :link_extra variable must be a string formatted as a valid HTML attribute/value pairs. That string will get inserted verbatim in the HTML of the link. (see more advanced details in the pagy_link_proc documentation)

Customizing the params

You may need to massage the params embedded in the URLs of the page links. You can do so by redefining the pagy_get_params sub-method in your helper. It will receive the params hash complete with the :page param and it should return a possibly modified version of it.

An example using except and merge!:

def pagy_get_params(params)
  params.except(:anything, :not, :useful).merge!(something: 'more useful')
end

You can also use the :param and : :anchor variables to add arbitrary params to the URLs of the pages. For example:

@pagy, @records = pagy(some_scope, params: {custom: 'param'}, anchor: '#your-anchor')

IMPORTANT: For performance reasons the :anchor string must include the "#".

Customizing the URL

When you need somethig more radical with the URL than just massaging the params, you should override the pagy_url_for right in your helper. The following are a couple of scenarios that would need that.

Enabling fancy-routes

The following is a Rails-specific alternative that supports fancy-routes (e.g. get 'your_route(/:page)' ... that produce paths like your_route/23 instead of your_route?page=23):

def pagy_url_for(page, pagy)
  params = request.query_parameters.merge(:only_path => true, pagy.vars[:page_param] => page )
  url_for(params)
end

Notice that this overridden method is quite slower than the original because it passes through the rails helpers. However that gets mitigated by the internal usage of pagy_link_proc which calls the method only once even in the presence of many pages.

POST with page links

You may need to POST a very complex search form that would generate an URL potentially too long to be handled by a browser, and your page links may need to use POST and not GET. In that case you can try this simple solution:

def pagy_url_for(page, _)
  page
end

That would produce links that look like e.g. <a href="2">2</a>. Then you can attach a javascript “click” event on the page links. When triggered, the href content (i.e. the page number) should get copied to a hidden "page" input and the form should be posted.

Customizing the item name

The pagy_info and the pagy_items_selector_js helpers use the “item”/”items” generic name in their output. You can change that by editing the values of the "pagy.item_name" i18n key in the dictionaray files that your app is using.

Besides you can also (dynamically) set the :i18n_key variable to let Pagy know where to lookup the item name in some dictionary file (instead looking it up in the default "pagy.item_name" key).

You have a few ways to do that:

1. you can override the pagy_get_vars method in your controller, adding the dynamically set :i18n_key. For example with ActiveRecord (mostly useful with the i18n extra or if you copy over the AR keys into the pagy dictionary):

    def pagy_get_vars(collection, vars)
      { count: ...,
        page: ...,
        i18n_key: "activerecord.models.#{collection.model_name.i18n_key}" }.merge!(vars)
    end
   

2. you can set the :i18n_key variable, either globally using the Pagy::VARS hash or per instance with the Pagy.new method or with the pagy controller method:

    # all the Pagy instances will have the default
    Pagy::VARS[:i18n_key] = 'activerecord.models.product'
   
    # or single Pagy instance
    @pagy, @record = pagy(my_scope, i18n_key: 'activerecord.models.product' )
   

Notice: The variables passed to a Pagy object have the precedence over the variables returned by the pagy_get_vars. The fastest way is passing a literal string to the pagy method, the most convenient way is using pagy_get_vars.

Paginate an Array

Please, use the array extra.

Paginate ActiveRecord collections

Pagy works out of the box with ActiveRecord collections. See also the arel for better performance of grouped ActiveRecord collections.

Paginate Ransack results

Ransack result returns an ActiveRecord collection, which can be paginated out of the box. For example:

@q = Person.ransack(params[:q])
@pagy, @people = pagy(@q.result)

Paginate Elasticsearch results

Pagy has a couple of extras for gems returning elasticsearch results: elasticsearch_rails and searchkick

Paginate pre-offsetted and pre-limited collections

With the other pagination gems you cannot paginate a subset of a collection that you got using offset and limit. With Pagy it is as simple as just adding the :outset variable, set to the initial offset. For example:

subset = Product.offset(100).limit(315)
@pagy, @paginated_subset = pagy(subset, outset: 100)

Assuming the :items default of 20, you will get the pages with the right records you are expecting. The first page from record 101 to 120 of the main collection, and the last page from 401 to 415 of the main collection. Besides the from and to attribute readers will correctly return the numbers relative to the subset that you are paginating, i.e. from 1 to 20 for the first page and from 301 to 315 for the last page.

Paginate non-ActiveRecord collections

The pagy_get_vars method works out of the box with ActiveRecord collections; for other collections (e.g. mongoid, etc.) you may need to override it in your controller, usually by simply removing the :all argument passed to count:

#count = collection.count(:all)
count = collection.count

Paginate Any Collection

Pagy doesn’t need to know anything about the kind of collection you paginate. It can paginate any collection, because every collection knows its count and has a way to extract a chunk of items given an offset and a limit. It does not matter if it is an Array or an ActiveRecord scope or something else: the simple mechanism is the same:

1. Create a Pagy object using the count of the collection to paginate
2. Get the page of items from the collection using pagy.offset and pagy.items

Here is an example with an array. (Please, notice that this is only a convenient example but you should use the array extra to paginate arrays).

# paginate an array
arr = (1..1000).to_a

# Create a Pagy object using the count of the collection to paginate
pagy = Pagy.new(count: arr.count, page: 2)
#=> #<Pagy:0x000055e39d8feef0 ... >

# Get the page of items using `pagy.offset` and `pagy.items`
paginated = arr[pagy.offset, pagy.items]
#=> [21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40]

This is basically what the pagy method included in your controller does for you in one go:

@pagy, @records = pagy(Product.some_scope)

Then of course, regardless the kind of collection, you can render the navigation links in your view:

<%== pagy_nav(@pagy) %>

See the Pagy::Backend source and the Pagy::Backend API documentation for more details.

Custom count for custom scopes

Your scope might become complex and the default pagy collection.count(:all) may not get the actual count. In that case you can get the right count with some custom statement, and pass it to pagy:

custom_scope = ...
custom_count = ...
@pagy, @records = pagy(custom_scope, count: custom_count)

Notice: pagy will efficiently skip its internal count query and will just use the passed :count variable

See also the arel for better performance of grouped ActiveRecord collections.

Using the pagy_nav* helpers

These helpers take the Pagy object and return the HTML string with the pagination links, which are wrapped in a nav tag and are ready to use in your view. For example:

<%== pagy_nav(@pagy) %>

Notice: the extras add a few other helpers that you can use the same way, in order to get added features (e.g. bootstrap compatibility, responsiveness, compact layouts, etc.)

Extra	Helpers
bootstrap	pagy_bootstrap_nav, pagy_bootstrap_nav_js, pagy_bootstrap_combo_nav_js
bulma	pagy_bulma_nav, pagy_bulma_nav_js, pagy_bulma_combo_nav_js
foundation	pagy_foundation_nav, pagy_foundation_nav_js, pagy_foundation_combo_nav_js
materialize	pagy_materialize_nav, pagy_materialize_nav_js, pagy_materialize_combo_nav_js
navs	pagy_nav_js, pagy_combo_nav_js
semantic	pagy_semantic_nav, pagy_semantic_nav_js, pagy_semantic_combo_nav_js
uikit	pagy_uikit_nav, pagy_uikit_nav_js, pagy_uikit_combo_nav_js

Helpers are the preferred choice (over templates) for their performance. If you need to override a pagy_nav* helper you can copy and paste it in your helper and edit it there. It is a simple concatenation of strings with a very simple logic.

Depending on the level of your overriding, you may want to read the Pagy::Frontend API documentation for complete control over your helpers.

Skipping single page navs

Unlike other gems, Pagy does not decide for you that the nav of a single page of results must not be rendered. You may want it rendered… or maybe you don’t. If you don’t… wrap it in a condition and use the pagy_nav* only if @pagy.pages > 1 is true. For example:

<%== pagy_nav(@pagy) if @pagy.pages > 1 %>

Skipping page=1 param

By default Pagy generates all the page links including the page param. If you want to remove the page=1 param from the first page link, just require the trim extra.

Using Templates

The pagy_nav* helpers are optimized for speed, and they are really fast. On the other hand editing a template might be easier when you have to customize the rendering, however every template system adds some inevitable overhead and it will be about 30-70% slower than using the related helper. That will still be dozens of times faster than the other gems, but… you should choose wisely.

Pagy provides the replacement templates for the pagy_nav, pagy_bootstrap_nav, pagy_bulma_nav, pagy_foundation_nav, and the pagy_uikit_nav helpers (available with the relative extras) in 3 flavors: erb, haml and slim.

They produce exactly the same output of the helpers, but since they are slower, using them wouldn’t make any sense unless you need to change something. In that case customize a copy in your app, then use it as any other template: just remember to pass the :pagy local set to the @pagy object. Here are the links to the sources to copy:

• pagy
  • nav.html.erb
  • nav.html.haml
  • nav.html.slim
• bootstrap
  • bootstrap_nav.html.erb
  • bootstrap_nav.html.haml
  • bootstrap_nav.html.slim
• bulma
  • bulma_nav.html.erb
  • bulma_nav.html.haml
  • bulma_nav.html.slim
• foundation
  • foundation_nav.html.erb
  • foundation_nav.html.haml
  • foundation_nav.html.slim
• uikit
  • uikit_nav.html.erb
  • uikit_nav.html.haml
  • uikit_nav.html.slim

If you need to try/compare an unmodified built-in template, you can render it right from the Pagy gem with:

<%== render file: Pagy.root.join('templates', 'nav.html.erb'), locals: {pagy: @pagy} %>
<%== render file: Pagy.root.join('templates', 'bootstrap_nav.html.erb'), locals: {pagy: @pagy} %>

You may want to read also the Pagy::Frontend API documentation for complete control over your templates.

Dealing with a slow collection COUNT(*)

Every pagination gem needs the collection count in order to calculate all the other variables involved in the pagination. If you use a storage system like any SQL DB, there is no way to paginate and provide a full nav system without executing an extra query to get the collection count. That is usually not a problem if your DB is well organized and maintained, but that may not be always the case.

Sometimes you may have to deal with some not very efficient legacy apps/DBs that you cannot totally control. In that case the extra count query may affect the performance of the app quite badly.

You have 2 possible solutions in order to improve the performance.

Caching the count

Depending on the nature of the app, a possible cheap solution would be caching the count of the collection, and Pagy makes that really simple.

Pagy gets the collection count through its pagy_get_vars method, so you can override it in your controller. Here is an example using the rails cache:

# in your controller: override the pagy_get_vars method so it will call your cache_count method
def pagy_get_vars(collection, vars)
  vars[:count] ||= cache_count(collection)
  vars[:page]  ||= params[ vars[:page_param] || Pagy::VARS[:page_param] ]
  vars
end

# add Rails.cache wrapper around the count call
def cache_count(collection)
  cache_key = "pagy-#{collection.model.name}:#{collection.to_sql}"
  Rails.cache.fetch(cache_key, expires_in: 20 * 60) do
   collection.count(:all)
  end
end

# in your model: reset the cache when the model changes (you may omit the callbacks if your DB is static)
after_create  { Rails.cache.delete_matched /^pagy-#{self.class.name}:/}
after_destroy { Rails.cache.delete_matched /^pagy-#{self.class.name}:/}

That may work very well with static (or almost static) DBs, where there is not much writing and mostly reading. Less so with more DB writing, and probably not particularly useful with a DB in constant change.

Using the arel extra

For better performance of grouped ActiveRecord collection counts, you may want to take a look at the arel.

Avoiding the count

When the count caching is not an option, you may want to use the countless extra, which totally avoid the need for a count query, still providing an acceptable subset of the full pagination features.

Paginate for API clients

When your app is a service that doesn’t need to serve any UI, but provides an API to some sort of client, you can serve the pagination metadata as HTTP headers added to your response.

In that case you don’t need the Pagy::Frontend nor any frontend extra. You should only require the headers extra and use its helpers to add the headers to your responses.

Paginate for Javascript Frameworks

If your app uses ruby as pure backend and some javascript frameworks as the frontend (e.g. Vue.js, react.js, …), then you may want to generate the whole pagination UI directly in javascript (with your own code or using some available component).

In that case you don’t need the Pagy::Frontend nor any frontend extra. You should only require the metadata extra and pass the pagination metadata in your JSON response.

Maximizing Performance

Here are some tips that will help chosing the best way to use Pagy, depending on your requirements and environment.

Consider the nav_js

If you need the classic pagination bar with links and info, then you have a couple of choices, depending on your environment:

• If you are on ruby 2.0+, add the oj gem to your gemfile and use any pagy*_nav_js helper (see Javascript). That uses client side rendering and it is faster and lighter than using any pagy*_nav helper (40x faster, 36x lighter and 1,410x more efficient than Kaminari). Notice: the oj gem is not a requirement but helps the performance when it is available.
• If you are on jruby (any version) or ruby 1.9.x, or you cannot install oj then use the pagy*_nav helper, which will give you the same performance of pagy v2.0 (33x faster; 26x lighter; 850x more efficient than Kaminari)

Consider the combo navs

If you don’t have strict requirements but still need to give the user total feedback and control on the page to display, then consider the pagy*_combo_nav_js helpers (faster and lighter when the oj gem is available). That gives you the best performance with nav info and UI (48x faster, 48x lighter and 2,270x more efficient than Kaminari) also saving real estate.

Consider the countless extra

If your requirements allow to use the countless extra (minimal or automatic UI) you can save one query per page, and drastically boost the efficiency eliminating the nav info and almost all the UI. Take a look at the examples in the support extra.

Preventing crawlers to follow look-alike links

The *_js helpers come with a JSON tag including a string that looks like an a link tag. It’s just a placeholder string used by pagy.js in order to create actual DOM elements links, but some crawlers are reportedly following it even if it is not a DOM element. That causes server side errors reported in your log.

You may want to prevent that by simply adding the following lines to your robots.txt file:

User-agent: *
Disallow: *__pagy_page__

Caveats: already indexed links may take a while to get purged by some search engine (i.e. you may still get some hits for a while even after you disallow them)

A quite drastic alternative to the robot.txt would be adding the following block to the config/initializers/rack_attack.rb (if you use the Rack Attack Middlewhare):

Rack::Attack.blocklist("block crawlers to follow pagy look-alike links") do |request|
  request.query_string.match /__pagy_page__/
end

but it would be quite an overkill if you plan to install it only for this purpose.

Handling Pagy::OverflowError exceptions

Pass an overflowing :page number and Pagy will raise a Pagy::OverflowError exception.

This often happens because users/clients paginate over the end of the record set or records go deleted and a user went to a stale page.

You can handle the exception by using the overflow extra which provides easy and ready to use solutions for a few common cases, or you can rescue the exception manually and do whatever fits better.

Here are a few options for manually handling the error in apps:

• Do nothing and let the page render a 500
• Rescue and render a 404

• Rescue and redirect to the last known page (Notice: the overflow extra provides the same behavior without redirecting)

   # in a controller
   rescue_from Pagy::OverflowError, with: :redirect_to_last_page
  
   private
  
   def redirect_to_last_page(exception)
     redirect_to url_for(page: exception.pagy.last), notice: "Page ##{params[:page]} is overflowing. Showing page #{exception.pagy.last} instead."
   end
  

WARNING: All Pagy exceptions are subclasses of ArgumentError, so if you need to rescue_from ArgumentError, ... along with rescue_from Pagy::OverflowError, ... then the Pagy::OverflowError line should go BEFORE the ArgumentError line or it will never get rescued.