How to count objects in repository... fast

[Gyro-Gearless]

Hi folks :-) go-git noob here...

I am playing around writing a little tool to backup remote git repositories, where backup just means "do a regularly git fetch".

Cloning repositories and doing the fetch goes pretty straightforward with go-git!

But now, i want to get some stats on how many (if any) update where fetched, so i had the idea to ask ChatGPT for a git count-objects surrogate, and came up with this:

(Idea is to do a count of objects before and after the fetch operation, and check the difference in counted objects)

func GetRepoStats(repo *git.Repository) (RepoStats, error) {

	counter := repoStatsImpl{}

	// Iterate over ALL raw objects in packfiles and loose storage
	iter, err := repo.Storer.IterEncodedObjects(plumbing.AnyObject)
	if err != nil {
		return nil, err
	}

	err = iter.ForEach(func(obj plumbing.EncodedObject) error {
		switch obj.Type() {
		case plumbing.CommitObject:
			counter.commits++
		case plumbing.TreeObject:
			counter.trees++
		case plumbing.BlobObject:
			counter.blobs++
		case plumbing.TagObject:
			counter.tags++
		default:
			counter.unknowns++
		}
		return nil
	})
	if err != nil {
		panic(err)
	}

	return &counter, nil
}

type repoStatsImpl struct {
	commits  int
	trees    int
	blobs    int
	tags     int
	unknowns int
}

// String implements RepoStats.
func (r *repoStatsImpl) String() string {
	return fmt.Sprintf("RepoStats:[Commits:%d,Trees:%d,Blobs:%d,Tags:%d,Unknown:%d]:%d", r.NumCommitObjects(), r.NumTreeObjects(), r.NumBlobObjects(), r.NumTagObjects(), r.NumUnknowns(), r.NumTotal())
}

This works and seems to give proper numbers, but... it is dog slow!

Even on a medium sized repository like this one: https://github.com/LMMS/lmms.git , the above calculation takes almost 5 minutes, whereas git count-objects -v almost immediately gives a result.

gyro@hal8999:~/work/g4g/butt$ go run ./cmd/count-objects/  /home/gyro/tmp/g4g/multimedia/music/production/daws/LMMS/
Scanning directory /home/gyro/tmp/g4g/multimedia/music/production/daws/LMMS/
 Stats: RepoStats:[Commits:28682,Trees:107094,Blobs:94607,Tags:4,Unknown:0]:230387
Time elapsed: 4m47.583518416s 

So i would be glad for any hint on how to make this faster - ultimately, i would just need the information if fetch picked up zero or more updates from remote...

Cheers, Gyro :-)