如何获取Sprockets生成资源的摘要？

Question

When using sprockets, I have a whole directory of .coffee files that transpile, merge together and become 'mobile_v2.js'. When this file gets served up a wonderful ETag is set on it with the SHA1 hash of the content. This allows cache-busting and everything else wonderful and good to work correctly. ie, if I modify any of the source .coffee files and reload the page, Sprockets will regenerate 'mobile_v2.js' and send it down to the client. If I don't change anything, the client will use it's cached copy of that resource. This works great.

Now I've got a scenario where a portion of the web app uses Backbone.js and the user can spend a really long time on one page without doing a full "refresh" ... ie, lots and lots of AJAX updates, but no new "...". Literally, the client could spend weeks on this page. Thus if I push an update to the JS/CSS resources, I need a way for the client to detect this and trigger a reload of the overall page. Doing a full reload is disruptive, so I don't want to do it more often than necessary. I also don't want to poll the server for these resources ... I have all these AJAX calls, so I can just piggy-back those calls with an additional response header. All that I can do. I just need help with one simple question:

Given a generated resource like 'mobile_v2.js', how do I query Sprockets to get the SHA1 digest for that file? (from my ruby code)

Answer 1

Not sure if I got you right, but if you're precompiling your assets, one way would be to extract the digest from the generated manifest.yml

manifest = YAML.load(File.open(Rails.root.join("public", "assets", "manifest.yml")))
digest = manifest["mobile_v2.js"].gsub("mobile_v2-", "").gsub(".js", "")

Answer 2

To find the sprockets digest of a generated asset:

Rails.application.assets.find_asset(asset_name).digest

In production, there's no sprockets(*), so we use the git revision (we deploy with the .git folder). Alternatively, I guess we could have read the manifest.yaml as suggested by DavidO.

Here's the full solution I ended up using:

def asset_digest(asset_name)
  if Rails.env == 'production'
    # Production
    Git.open(Rails.root).object('HEAD').sha
  else
    # Development / Test
    Rails.application.assets.find_asset(asset_name).digest
  end 
end

(*): Actually, in production, the sprockets line "worked", but it did so by recompiling all of the assets instead of reading the precompiled asset.