brancher: do not get tree if there's no revs

[skshetry]

From #3207 (comment)

When if it's a dvc only repo, brancher was failing as NoSCM does not have resolve_rev.

$ dvc init --no-scm
$ python -c '
from dvc.repo import Repo
repo = Repo(".")
print(list(repo.brancher(all_branches=True, all_commits=True, all_tags=True)))
'
Traceback (most recent call last):
  File "<string>", line 4, in <module>
  File "/home/saugat/repos/iterative/dvc/dvc/repo/brancher.py", line 45, in brancher
    for sha, names in group_by(scm.resolve_rev, revs).items():
AttributeError: 'NoSCM' object has no attribute 'resolve_rev'

• ❗ Have you followed the guidelines in the Contributing to DVC list?

• 📖 Check this box if this PR does not require documentation updates, or if it does and you have created a separate PR in dvc.org with such updates (or at least opened an issue about it in that repo). Please link below to your PR (or issue) in the dvc.org repo.

• ❌ Have you checked DeepSource, CodeClimate, and other sanity checks below? We consider their findings recommendatory and don't expect everything to be addressed. Please review them carefully and fix those that actually improve code or fix bugs.

Thank you for the contribution - we'll try to review it as soon as possible. 🙏

[efiop]

Nice catch! Thanks!

[ghost]

this should be caught by:

    if not any([revs, all_branches, all_tags, all_commits]):
        yield ""
        return

🤔

[ghost]

ohhh, but I saw that you were doing all_*=True 🙂

[ghost]

Wait, @skshetry , why didn't you solve it by using if not scm: return ?

[ghost]

As far as I understand, brancher doesn't have anything to do when NoSCM is used.

So, it would be:

    scm = self.scm

    if not any([revs, all_branches, all_tags, all_commits]) or not scm:
        yield ""
        return

I guess that, since the end result is the same, it doesn't matter

[skshetry]

scm is always there. If it's a non-git repo, it will be NoSCM.
Honestly, I didn't really feel confident around brancher() (this is an uncovered corner). I really just added here because of the finally statement here that resets the self.tree.

[efiop]

Which doesn't make your point wrong, I'm just saying.

[ghost]

@efiop , I don't get why revs=["something"] is intentionally breaking it but no all_commits=True.

[efiop]

@MrOutis because it expects to receive revs that are valid in the scm that we are operating in and there are no valid revs in NoSCM. And all_commits=True asks scm for list of all commits, which are [] for NoSCM. I agree that brancher could be improved to handle those cases more smoothly, but this fix does get the job done in this case.

[skshetry]

As far as my understanding goes, the NoSCM case and working tree should not even be the concern of the brancher().

[skshetry]

Looks like someone has been here and changed things, and forgot to update the docstrings.