Fix digest images being lost on load of hauls (Signed).

[KaminFay]

Please check below, if the PR fulfills these requirements:

• Commit(s) and code follow the repositories guidelines.
• Test(s) have been added or updated to support these change(s).
• Doc(s) have been added or updated to support these change(s).

Associated Links:

• [BUG] Images imported with digests fail to re-appear on the other archival process #250

Types of Changes:

• Bugfix

Proposed Changes:

Note:

Had to republish the PR after fixing my laptops commit signing setup.

• Current when the image reference is taken by the Pusher() method and split, it is done so based on the first instance of the @ symbol:

var baseRef, hash string
parts := strings.SplitN(ref, "@", 2)
baseRef = parts[0]
if len(parts) > 1 {
    hash = parts[1]
}

• Proposed way of splitting:

	var baseRef, hash string

	parts := strings.Split(ref, "@")
	baseRef = strings.Join(parts[:len(parts)-1], "@")
	if len(parts) > 1 {
		hash = parts[len(parts)-1]
	}

	return baseRef, hash

• In reality that will cause the hash to appear with a lot of extra junk if there is more than one @ symbol in the reference string, like there is with digests.... Ex:

sha256:95b01e2e9ab0702ce2f1a8f05f90e6408fd1f4b5e5006c6088ba5a864ed42064-library/nginx@sha256:95b01e2e9ab0702ce2f1a8f05f90e6408fd1f4b5e5006c6088ba5a864ed42064-dev.cosignproject.cosign/imageIndex@sha256:95b01e2e9ab0702ce2f1a8f05f90e6408fd1f4b5e5006c6088ba5a864ed42064"

• The proposed solution will instead filter on the last @ symbol instead of the first. Resulting in a proper hash being pulled out.
• A long with this the splitting on the hash and base reference string has been pulled out into its own function so that it can be tested using the newly included testing methods. Just to make sure digests are not broken again in the future.
  Verification/Testing of Changes:

• The change can be verified by following the steps in the bug report and seeing that we now have the correct index on unpacking and the info command shows everything is where it should be.

[dweomer]

As noted here I would very much prefer we try a more standard approach and try to leverage https://github.com/distribution/reference for correctness if not concision.

[zackbradys]

Thanks for reviewing and approving @dweomer... will test this tonight before merging.

@KaminFay do you notice any issues?

[KaminFay]

Thanks for reviewing and approving @dweomer... will test this tonight before merging.

@KaminFay do you notice any issues?

Nope no issues seen on my side

[zackbradys]

Glad to hear it! @KaminFay

[zackbradys]

I was finally able to run a few tests and LGTM!!