moocode / asset_id Goto Github PK

I'm preparing mime-types 2.0 for release, and it has some breaking API changes (not for most uses, but some esoteric features). The most important API change is that mime-types 2.0 no longer supports Ruby 1.8.

If this matters, the gemspec needs to be changed from >= 1.16 to ~> 1.16.

I use a sprite plugin that appends the timestamp to the url in css:

url(/images/logo.png?194819891)

So if you just add url = url.split("?")[0], to the following method, it works great!

def replace_css_images!(options={})
  options[:prefix] ||= ''
  # adapted from https://github.com/blakink/asset_id
  data.gsub! /url\((?:"([^"]*)"|'([^']*)'|([^)]*))\)/mi do |match|
    begin
      # $1 is the double quoted string, $2 is single quoted, $3 is no quotes
      uri = ($1 || $2 || $3).to_s.strip
      uri.gsub!(/^\.\.\//, '/')

      # if the uri appears to begin with a protocol then the asset isn't on the local filesystem
      if uri =~ /[a-z]+:\/\//i
        "url(#{uri})"
      else
        url = url.split("?")[0]
        asset = Asset.new(uri)
        # TODO: Check the referenced asset is in the asset_paths
        puts "  - Changing CSS URI #{uri} to #{options[:prefix]}#{asset.fingerprint}" if @@debug
        "url(#{options[:prefix]}#{asset.fingerprint})"
      end
    rescue Errno::ENOENT => e
      puts "  - Warning: #{uri} not found" if @@debug
      "url(#{uri})"
    end
  end
end

It would be ideal if one could specify the location of the yml file holding the credentials for S3 to keep things DRY.

Hey, have you any good way to set a file path e.g. in CSS with url()...
I can hardcode url(http://static.host.com/images/pic.jpg) but then in development mode, maybe when I'm offline, I have the trouble.
And I think it's a stupid way to hardcode the full host address.
Maybe I will change the domain in the future and so on.

Daniel

Since asset_id inserts an md5 hash into the filename of each asset uploaded, wouldn't it be faster to check on s3 for the presence of the file before reuploading every resource every time?

Hi. first of all, thanks for putting asset_id together, it's very handy.

My project uses compass/sass, so I need to modify the output of the compass image-url method. compass has it's own asset_host specification system, it doesn't cleverly inherit from rails or anything.

Was just wondering if you've used asset_id with compass and how you achieved that. If not, I'll let you know what I put together.

We have two buckets for production and staging.
Once we uploaded to staging then we can't upload to production without deleting log file.