CarrierWave sets wrong content type on XLSX files

[resistorsoftware]

My incoming upload has the right content type. A before store action shows the right content type. But the actual stored file on S3 has the wrong content type!! How to correct that?

incoming file:
2016-08-05T15:02:16.358472+00:00 app[web.1]: new_rates.xlsx
2016-08-05T15:02:16.358473+00:00 app[web.1]: file
2016-08-05T15:02:16.358474+00:00 app[web.1]: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet

2016-08-05T15:02:16.401995+00:00 app[web.1]: BEFORE STORE: {:filename=>"new_rates.xlsx", :type=>"application/vnd.openxmlformats-officedocument.spreadsheetml.sheet", :name=>"file", :tempfile=>#<Tempfile:/tmp/RackMultipart20160805-6-cdv09y.xlsx>, :head=>"Content-Disposition: form-data; name=\"file\"; filename=\"new_rates.xlsx\"\r\nContent-Type: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet\r\n"}
2016-08-05T15:02:16.707430+00:00 app[web.1]: Uploaded file content type: application/zip
2016-08-05T15:02:16.707446+00:00 app[web.1]: File Type is:application/zip

[bemosior]

@resistorsoftware here's an extremely terrible hack I'm using in the meantime:

class AttachmentUploader < CarrierWave::Uploader::Base

# ...

  process :override_content_type

  def override_content_type
    if File.extname(file.file).delete('.').to_sym == :xlsx
      file.content_type='application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'
    end
  end

#...

end

[bemosior]

If I'm following everything correctly...

The documentation here states:

As of v0.11.0, the mime-types gem is a runtime dependency and the content type is set automatically. You no longer need to do this manually

The mime-types gem appears to be: https://github.com/mime-types/ruby-mime-types

And it's seems these sorts of files are not accommodated. I could not find XLSX or similar files in the supported types, so I believe this to be an issue in that upstream dependency.

[bemosior]

I became hopeful when I read the Contributing documentation:

The mime-types registry is no longer contained in mime-types, but in mime-types-data. Please see that project for contributions there.

The latest versions of mime-types-data appear to support XLSX and others, so I'm not sure what's happening here.

[bemosior]

I forked one of @mehlah's content-type test scripts, hacked in a quick content-type check on an XLSX doc, and it correctly detected application/vnd.openxmlformats-officedocument.spreadsheetml.sheet: https://gist.github.com/bemosior/d341b2c4423f7b8347b0267387e15eb1

[bemosior]

@resistorsoftware: In my case, I've determined that there's a disconnect between the content-type sent by our front-end and that actually delivered to Carrierwave. It's definitely an issue somewhere within our own code. I suggest getting a handle on all the code layers in your control between upload in the browser and the Carrierwave library and verifying the content-type perceived in each.

[bemosior]

The resolution to my issue ended up looking like this:

def attachment
  #Added this line to resolve what turned out to be an attribute mapping issue:
  file[:content_type] = file[:type]

  @attachment ||= Attachment.new(file: file)
end

Hope this helps.

[resistorsoftware]

Thanks so much for the explanations. I validated using Pry that the gem should've been able to detect the proper content type using the extension as is, which troubled me. Why was the content type mangled? So at that point I made a hack on the result, and used a really dumb update on the S3 file to force its content type to be correct.

I will follow through this thread and see if I can get CarrierWave to work thus eliminating the crazy S3 hack!

Thanks again!

[arseny-emchik]

The same problems for me

[p8]

Mimemagic comes with extra magic you can overlay on top of the defaults to correctly detect these file types. Enable it like this:

    require 'mimemagic/overlay'
    MimeMagic.by_magic(File.open('test.xlsx'))

See: https://github.com/minad/mimemagic#extra-magic-overlay

[mshibuya]

#1934 has made MimeMagic as default way to detect MIME types, so this should be no longer an issue.

[andyrue]

I recently upgraded from 1.3.1 to the latest 2.1.0 and I am now experiencing this same problem with XLSX and DOCX files. The uploader detects things properly, but after saving the file the content_type is application/zip. Rolling back to 1.3.1 fixes it.

[nathancolgate]

We experienced the same issue and fix as @andyrue on a Rails 5.2 app.

[tmr08c]

It looks like this may have been fixed again in #2447 which was included in the 2.2.0 release 🎉

[phansch]

We are running into a similar problem with .apk files, which should be application/vnd.android.package-archive but are interpreted as application/zip. Is there a way to override this from our side, or does that require a fix here?

[phansch]

Currently using this temporary workaround for setting the content type of the file manually:

class SomeUploader < CarrierWave::Uploader::Base
  process :apk_content_type

  def apk_content_type
    if current_path.end_with?('.apk')
      file.content_type = 'application/vnd.android.package-archive'
    end
  end