Rake Pipeline filter for syncing already-published static content on production with the generated local file via Nokogiri and css selectors

Assetfile

require "rake-pipeline-web-filters"
require "./remote-replace-filter"

# specify output, other configuration

input "app" do
  match "slim/whatever.slim" do
    tilt { |input| input.sub(/\.slim$/, '.html') }

    copy "index.html"

    # If you don't want to perform the remote query while using
    # `rakep server`, you can append "defined?(::Rake::Pipeline::Server)"
    # to this line.
    # NOTE: make sure this line comes after the last filter to change
    # the output path (e.g. copy "index.html" from above) so that the filter
    # can determine the remote location of the already published file.
    remote_replace "http://www.yourstaticsite.com", ".content" 
  end

end

remote-replace-filter.rb

require "rake-pipeline"
require "rake-pipeline-web-filters"
require "pathname"
require 'net/http'
require 'uri'

# A filter that uses Nokogiri to replace input content
# with remote content.
# Use case: you deploy a static website that is maintained
# by someone editing HTML and pushing via FTP, and you need
# to make a change to non-content portion of the page. So that
# you don't overwrite the client-generated content, you can
# use this filter to specify the CSS selectors that will be
# copied from the remote server into your generated output.
class RemoteReplaceFilter < Rake::Pipeline::Filter

  class MissingLocalContent < StandardError; end

  include Rake::Pipeline::Web::Filters::FilterWithDependencies

  # @param [String] host (required) a valid http address that'll
  #                 be used in conjunction with the relative path
  #                 of the input file to resolve the remote file
  #
  #   selectors: CSS selectors
  # @param [Proc] block the output name generator block
  def initialize(host, *selectors, &block)
    super(&block)

    @host = Pathname.new host
    raise ArgumentError, "at least one CSS selector required" if selectors.empty? || selectors.first.empty?
    @selectors = selectors
  end

  # The body of the filter. Query the remote equivalent of
  # each input file and swap out the local file's selected content
  # with the remote file's selected conttent.
  #
  # @param [Array] inputs an Array of FileWrapper objects.
  # @param [FileWrapper] output a FileWrapper object
  def generate_output(inputs, output)
    inputs.each do |input|
      remote_url = @host.join(input.path).to_s

      remote_doc = Nokogiri::HTML(Net::HTTP.get(URI.parse(remote_url)))
      local_doc  = Nokogiri::HTML(input.read)

      # Replace content for each selector
      @selectors.each do |sel|
        remote_element = remote_doc.at_css(sel)
        remote_content = remote_element ? remote_element.content : "Couldn't find remote content for selector: #{sel}"
        local_element = local_doc.at_css(sel)
        raise MissingLocalContent, "Couldn't find local element in #{input.path} via selector: #{sel}" unless local_element
        local_element.content = remote_content
      end

      # Write the replaced HTML to output
      output.write local_doc
    end
  end

  def external_dependencies
    [ "nokogiri" ]
  end
end

# Add convenience function so that all you have to put in your
# Asset file is `remote_replace` vs filter(RemoteReplaceFilter)...
Rake::Pipeline::DSL::PipelineDSL.module_eval do
  # Add a new {RemoteReplaceFilter} to the pipeline.
  # @see RemoteReplaceFilter#initialize
  def remote_replace(*args, &block)
    filter(RemoteReplaceFilter, *args, &block)
  end
end