Creating an S3 signed URL without downloading the object

[benbc]

I'm using fog to generate signed URLs for S3, like this:

s3.directories.get(BUCKET).files.get(object).url(expiry)

This causes fog to download the entire object, which I don't need to do.

Is there any way to create the signed URL without downloading the object?

[benmanns]

You can use head to avoid downloading the whole object (this does a HEAD request, which pulls headers but not the file content). If you don't care about pulling headers or checking that the object exists, you can use new to build a file object, which you can use to create a URL without any requests S3.

Using head to validate object existence:

s3.directories.get(BUCKET).files.head(object).url(expiry)

Without touching S3:

s3.directories.new(:key => BUCKET).files.new(:key => object).url(expiry)

[benbc]

Wonderful. Thank you.

[geemus]

@benmanns thanks for covering this, spot on info.

[will]

Woah, thanks to mark fine for linking this issue. My app was surprise downloading huge things just to make a url, I had no idea.

[geemus]

@will - fwiw, there is another way which avoids the head request also (if you can reasonably assume that the file exists). Happy to discuss or help out with any S3 stuff you might have going on, I have used it a lot, probably too much.

[tischler]

@geemus this page comes up fairly easy when googling for signed URLs. if you know a better way, you should add a link to this thread... cheers!

[geemus]

Hmm, hard to remember now. I think basically you can just do the second part above (with new calls), but skip the head step. If the file doesn't really exist it would be invalid, but if you have reason to believe it does, it saves you the round-trip. I think that is probably what I meant, but, has been a while. Hope that helps!