cms_asset_import.rb

# script to import cms assets from those hosted on cms.housetrip.com and
# upload them as cms asset objects to cloudfront, also updating all
# references to the images in cms_page_contents markdown for all locales
# only the about us and press corner pages contain assets

# create a tmp dir for this script to keep downloaded assets in
`rm -rf tmp/cms_asset_import`
`mkdir -p tmp/cms_asset_import`

# helper methods

def original_local_file_path(url)
  uri      = URI.parse(url)
  filename = uri.path.split('/')[-1]
  digest   = Digest::MD5.hexdigest(uri.path)[0..5]
  "tmp/cms_asset_import/#{digest}-#{filename}"
end

def download_image(url)
  local_path = original_local_file_path(url)
  uri = URI.parse(url)
  response = Net::HTTP.start(uri.host, uri.port) {|http|
    http.get(uri.path)
  }
  case response
  when Net::HTTPSuccess
    image_data = response.body
    return nil if image_data.nil? || image_data.size == 0
    filename = uri.path.split('/')[-1]
    open(local_path, "wb") { |file| file.write(image_data) }
    return true
  end
  false
end

cms_pages  = CmsPage.find(:all, :conditions => ['identity IN (?)', ['about-us', 'press-corner']])

# scan pages for asset urls
asset_urls = []
cms_pages.each do |cms_page|
  puts "looking for assets on; '#{cms_page.name}'"
  cms_page.cms_page_contents.each do |cms_page_content|
    # find all linked assets from cms.housetrip.com in content
    urls = cms_page_content.body.scan(/(http:\/\/cms\.housetrip\.com.*(png|jpg|pdf|gif|jpeg|PDF))/)
    urls.collect! {|url| url[0] if url[0] =~ /http/ }
    break unless urls.present?
    puts " - #{cms_page_content.locale}  (#{urls.length} assets)"
    asset_urls += urls
  end
end
asset_urls.uniq!

# download assets and build replacements array
replacements = []
puts "downloading #{asset_urls.length} assets"
asset_urls.each do |asset_url|
  # download the assets
  if download_image(asset_url)
    # upload the asset
    if new_cms_asset = CmsAsset.create(:asset => File.open(original_local_file_path(asset_url)))
      # set replacement url to gsub
      replacements << { :original_url => asset_url, :new_url => new_cms_asset.asset.url }
    else
      puts "couldn't create asset for #{asset_url}"
    end
  end
end

# update all references within cms_page_content contents
cms_pages.each do |cms_page|
  puts "replacing asset urls on; '#{cms_page.name}'"
  cms_page.cms_page_contents.each do |cms_page_content|
    new_body = cms_page_content.body
    replacements.each do |replacement|
      new_body.gsub!(replacement[:original_url], replacement[:new_url])
    end
    cms_page_content.reload
    cms_page_content.body = new_body
    if cms_page_content.save!
      puts " - #{cms_page_content.locale} done!"
    end
  end
end