Fast HTML helpers using Nokogiri—pick and choose what you want

nokogiri_html_helpers.gemspec

Gem::Specification.new do |s|
  s.name = 'nokogiri_html_helpers'
  s.version = '0.1.4'
  s.platform = Gem::Platform::RUBY
  s.author = 'Samuel Cochran'
  s.email = 'sj26@sj26.com'
  s.homepage = 'https://gist.github.com/sj26/7018858'
  s.summary = 'Fast HTML helpers using Nokogiri'
  s.description = s.summary
  s.license = 'MIT'

  s.files = ['nokogiri_html_helpers.rb', 'strip_empty_block_html_helper.rb', 'truncate_html_helper.rb']
  s.test_files = ['strip_empty_block_html_helper_spec.rb', 'truncate_html_helper_spec.rb']
  s.require_path  = '.'

  s.add_dependency('nokogiri', ["~> 1.6.0"])
  s.add_development_dependency('rspec', ["~> 2.0"])
end

nokogiri_html_helpers.rb

module NokogiriHtmlHelpers
  class Engine < Rails::Engine
    def find_root_with_flag(*); File.expand_path("..", __FILE__); end
    self.paths["app/helpers"] = ""
  end if defined? Rails
end

strip_empty_block_html_helper.rb

module StripEmptyBlockHtmlHelper
  def strip_empty_block_html(html)
    Nokogiri::HTML::DocumentFragment.parse(html).tap do |fragment|
      fragment.traverse do |node|
        if node.element? and node.description and node.description.block? and node.content.empty?
          node.remove
        end
      end
    end.to_html
  end
end

strip_empty_block_html_helper_spec.rb

require "./strip_empty_block_html_helper"

describe StripEmptyBlockHtmlHelper do
  include StripEmptyBlockHtmlHelper

  specify "It should remove empty block-level elements" do
    strip_empty_block_html("<p>Thing</p><p><span><!-- img was here --></span></p><p>Stuff</p>").should == "<p>Thing</p><p>Stuff</p>"
  end
end

truncate_html_helper.rb

require "nokogiri"

module TruncateHtmlHelper
  def truncate_html(html, options={}, &block)
    if html
      length = options.fetch(:length, 30)
      omission = options.fetch(:omission, "…")

      omission &&= Nokogiri::HTML::DocumentFragment.parse(omission)

      remaining = length
      remaining -= omission.content.length if omission

      fragment = Nokogiri::HTML::DocumentFragment.parse(html)
      fragment.traverse do |node|
        if node.text? or node.cdata?
          remaining -= node.content.length
          if remaining < 0
            node.content = node.content[0...remaining]
          end
          if node.content.empty?
            node.remove
          end
        elsif node.element? and remaining <= 0 and not node.children.any?
          node.remove
        end
      end

      # If there's a block, use it instead of the omission
      if block_given? && remaining < 0
        omission = Nokogiri::HTML::DocumentFragment.parse(capture(&block))
      end

      # Add the omission only if we have truncated
      if omission && remaining < 0
        last_child = fragment.children.last
        # If the last element is block, level, tuck the omission inside
        if last_child.element? and last_child.description and last_child.description.block?
          last_child << omission
        else
          fragment << omission
        end
      end

      fragment.to_html
    end
  end
end

truncate_html_helper_spec.rb

require "./truncate_html_helper"

describe TruncateHtmlHelper do
  include TruncateHtmlHelper

  specify "basic truncation inside block-level element" do
    truncate_html("<p>A <strong>thing</strong> which gets truncated.</p><p>With trailing elements</p>", length: 8).should == "<p>A <strong>thing</strong>…</p>"
  end

  specify "empty elements are preserved, but danglers are removed" do
    truncate_html("<p>And<br>line<br>breaks<br>are<br>preserved.</p>", length: 8).should == "<p>And<br>line…</p>"
  end

  specify "omissions are outside inline elements" do
    truncate_html("<strong>Thing</strong> and <em>stuff</em>", length: 6).should == "<strong>Thing</strong>…"
  end

  specify "default length is 30" do
    truncate_html("a" * 50).should == "a" * 29 << "…"
  end

  specify "alternate omission text" do
    truncate_html("<div>a<em>b</em>cdef</div>", length: 6, omission: "...").should == "<div>a<em>b</em>c...</div>"
  end
end