git blame panic: contents and commits have different length

[williamfzc]

func TestAA(t *testing.T) {
	abs, _ := filepath.Abs("F:\\workspace\\github\\go-git")
	repo, err := git.PlainOpen(abs)
	if err != nil {
		panic(err)
	}
	head, err := repo.Head()
	if err != nil {
		panic(err)
	}
	targetCommit, err := repo.CommitObject(head.Hash())
	if err != nil {
		panic(err)
	}

	_, err = git.Blame(targetCommit, "remote.go")
	if err != nil {
                // PANIC!!!
		panic(err)
	}
}

panic happened when calling git.Blame

panic: contents and commits have different length [recovered]
	panic: contents and commits have different length

It will happened also in work_tree_test.go.
Looks like it can not handle extra empty line at the end of file.

In git blame:

But actually:

Tested on Windows and WSL.

guess

func newLines(contents []string, commits []*object.Commit) ([]*Line, error) {
	lcontents := len(contents)
	lcommits := len(commits)

	if lcontents != lcommits {
		fmt.Printf("c vs c: %d, %d", lcontents, lcommits)
...

Output:

=== RUN   TestAA
c vs c: 1430, 1397--- FAIL: TestAA (1.97s)

lcommits seems not correct.

[williamfzc]

I printed all its revs in go-git and found the last one is 0dcebfb on Dec 05 2021.
But it's not correct.

The real last one is f0b111a on Dec 10 2021.
I think this problem happened when a commit has multi parents.

[williamfzc]

ok it causes by:

go-git/references.go

Line 105 in 08cffa1

// TODO: detect merges that had a conflict, because they must be

When it has multi parents, itself will not be added to result list, which causes some commits lost.

williamfzc@188480d

This fix looks fine.

[tarihub]

I also meet this issue too, any official fix for this ?

[tarihub]

ok it causes by:

go-git/references.go

Line 105 in 08cffa1

// TODO: detect merges that had a conflict, because they must be

When it has multi parents, itself will not be added to result list, which causes some commits lost.

williamfzc@188480d

This fix looks fine.

This fix me too~
Thanks

[pjbgf]

@williamfzc would you be keen to propose your changes as a PR? If so, tag me and I will review it for you.

[williamfzc]

@pjbgf

Cool I have submitted one:
#751