Before and after refactoring of an asset checker - that looks at all assets in a rails project, and finds the ones that appear to be unreferenced.

asset_optimizer_after.rb

require 'rubygems'
require 'find'
require 'pp'

class UnusedImagesFinder

  def find_unused_assets
    puts "Lists unused files based upon explicit programatic checks. Requires manual confirmation before deletion."
    puts "------------------ Building css from sass ------------------"
    system 'rake build_static:stylesheets'

    puts "------------------ Finding assets and references ------------------"
    all_assets = find_all_asset_files
    used_assets = find_all_asset_references
    delta = all_assets.values.flatten - used_assets.keys
    puts "----------------- #{delta.size} Files that might not be used -----------------------"
    pp delta

    delta = used_assets.keys - all_assets.values.flatten
    puts "------------ #{delta.size} Files referenced but not at computed location ---------------"
    pp delta
  end

  def find_all_asset_files
    ignored_dirpaths = %w(./public/system ./public/fonts ./public/javascripts)
    ignored_extensions = %w(html htm txt css) << ''
    asset_files = {}
    Find.find('./public') do |f|
      extension = extension(f)
      next if File.directory?(f) || ignored_extensions.member?(extension) || ignored_dirpaths.any?{|prefix| File.dirname(f).start_with?(prefix)}
      (asset_files[extension] || asset_files[extension] = []) << f
    end
    asset_files
  end

  def find_all_asset_references
    checkers = {'erb' => :add_image_matches, 'css' => :add_css_matches}
    refs = {}
    messages = []
    %w(./app/views public/stylesheets).each do |base|
      Find.find(base) do |file|
        next if File.directory?(file)
        lines = File.readlines(file)
        lines.each_with_index do |line, line_number|
          checker = checkers[extension(file)]
          send(checker, file, refs, line, line_number) if checker
        end
      end
    end
    refs
  end

  private

  def add_image_matches(file, refs, line, line_number)
    add_matches(/img.*src=['"]([^'"]+)['"]/, file, refs, line, line_number) {|match| "./public/images/#{match[1]}"}
    add_matches(/<%=[ ]*image_tag[ ]+["']([^'"]+)['"]/, file, refs, line, line_number) {|match| "./public/images/#{match[1]}"}
  end

  def add_css_matches(file, refs, line, line_number)
    add_matches(/url\([\s]*["']?([^\s'"\)]+)['"]?[\s]*\)/, file, refs, line, line_number) {|match| match[1].gsub('../images', './public/images')}
  end

  def add_matches(regex, file, refs, line, line_number)
    match = line.match(regex)
    if match
      image = yield match
      puts "#{file}:#{line_number+1} #{image}\t\t#{line}"
      (refs[image] || refs[image] = []) << file
    end
  end

  def extension(file)
    File.extname(file).gsub('.', '')
  end

end

#UnusedImagesFinder.new.find_all_asset_files
#UnusedImagesFinder.new.find_all_asset_references
UnusedImagesFinder.new.find_unused_assets

asset_optimizer_before.rb

require 'rubygems'
require 'find'
require 'pp'

class UnusedImagesFinder

  def find_unused_assets
    puts "Lists unused files based upon explicit programatic checks. Requires manual confirmation before deletion."
    puts "------------------ Build css from sass ------------------"
    system 'rake build_static:stylesheets'

    all_assets = find_all_asset_files
    used_assets = find_all_asset_references
    delta = all_assets.values.flatten - used_assets.keys
    puts "----------------- #{delta.size} Files that might not be used -----------------------"
    pp delta

    delta = used_assets.keys - all_assets.values.flatten
    puts "------------ #{delta.size} Files referenced but not at computed location ---------------"
    pp delta
  end

  def find_all_asset_files
    ignored_dirpaths = %w(./public/system ./public/fonts ./public/javascripts)
    ignored_extensions = %w(html htm txt css) << ''
    asset_files = {}
    Find.find('./public') do |f|
      extension = extension(f)
      next if File.directory?(f) || ignored_extensions.member?(extension) || ignored_dirpaths.any?{|prefix| File.dirname(f).start_with?(prefix)}
      (asset_files[extension] || asset_files[extension] = []) << f
    end
    asset_files
  end

  def find_all_asset_references
    refs = {}
    messages = []
    %w(./app/views public/stylesheets).each do |base|
      Find.find(base) do |f|
        next if File.directory?(f)
        lines = File.readlines(f)
        extension = extension(f)
        lines.each_with_index do |line, line_number|
          add_image_matches(f, refs, line, line_number, messages) if extension == 'erb'
          add_css_matches(f, refs, line, line_number, messages) if extension == 'css'
        end
      end
    end
    refs
  end

  private

  def add_image_matches(file, refs, line, line_number, messages)
    matched_img = line.match(/img.*src=['"]([^'"]+)['"]/)
    if matched_img
      image = "./public/images/#{matched_img[1]}"
      messages << "#{file}:#{line_number+1} #{image}\t#{line}"
      (refs[image] || refs[image] = []) << file
    end
    matched_image_tag = line.match(/<%=[ ]*image_tag[ ]+["']([^'"]+)['"]/)
    if matched_image_tag
      image = "./public/images/#{matched_image_tag[1]}"
      messages << "#{file}:#{line_number+1} #{image}\t#{line}"
      (refs[image] || refs[image] = []) << file
    end
  end

  def add_css_matches(file, refs, line, line_number, messages)
    matched_url = line.match(/url\([\s]*["']?([^\s'"\)]+)['"]?[\s]*\)/)
    if matched_url
      image = matched_url[1].gsub('../images', './public/images')
      messages << "#{file}:#{line_number+1} #{image}\t#{line}"
      (refs[image] || refs[image] = []) << file
    end
  end

  def extension(file)
    File.extname(file).gsub('.', '')
  end

end

#UnusedImagesFinder.new.find_all_asset_files
#UnusedImagesFinder.new.find_all_asset_references
UnusedImagesFinder.new.find_unused_assets