Open mfdataset enchancement

[pratiman-91]

• Closes better handling of invalid files in open_mfdataset #6736
• User visible changes (including notable bug fixes) are documented in whats-new.rst

Added new argument in open_mfdataset to better handle the invalid files.

errors : {'ignore', 'raise', 'warn'}, default 'raise'
        - If 'raise', then invalid dataset will raise an exception.
        - If 'ignore', then invalid dataset will be ignored.
        - If 'warn', then a warning will be issued for each invalid dataset.

[welcome]

Thank you for opening this pull request! It may take us a few days to respond here, so thank you for being patient.
If you have questions, some answers may be found in our contributing guidelines.

[max-sixty]

I'm not the expert, but this looks reasonable! Any other thoughts?

Assuming no one thinks it's a bad idea, we would need tests.

[headtr1ck]

I think it is a good idea.

But the way it is implemented here seems overly complicated and repetitive.
I would suggest to revert the logic: first build up the list wrapped in a single try and then handle the three cases in the except block.

[headtr1ck]

Almost there.

Also, we should add tests for this.

[pratiman-91]

@headtr1ck Thanks for the suggestions. I have added two tests (ignore and warn). Also, while testing, I found that a new argument broke combine="nested" due to invalid ids. I have now modified it to reflect the correct ids, and it is passing the tests. Please review the tests and the latest version.

[pratiman-91]

Hi @headtr1ck, I have been thinking about the handling of ids. Current version looks like a patch work (I am not happy with it.). I think we can create ids after removing all the invalid datasets from path1d within the combine==nested block. Please let me know what do you think.
Thanks!

[pratiman-91]

@max-sixty Can you please go through the PR. Thanks!

[max-sixty]

I'm admittedly much less familiar with this section of the code. nothing seems wrong though!

I think we should bias towards merging, so if no one has concerns then I'd vote to merge

could we fix the errors in the docs?

[pratiman-91]

It seems like one test failed test_sparse_dask_dataset_repr (xarray.tests.test_sparse.TestSparseDataArrayAndDataset) . It is not related to this PR.

[pratiman-91]

can we fix the errors in the docs?

I just rebased the branch. Let us see if this resolves the docs. Ahh.. found the problem.

[pratiman-91]

@max-sixty all tests passed.

[kmuehlbauer]

@pratiman-91 Thanks for sticking with this. This seams all reasonable to me. I've added a couple comments and suggestions.

For the tests, I think it would be great to also test for "raise".

[pratiman-91]

@kmuehlbauer Thank you for the suggestions and for fixing the typos. I have made changes based on your suggestions. Please check.

For the tests, I did not write a test for "raise" since it is the default and already being tested by various other tests. If you think it would be beneficial to add another test, then I can write it.

[kmuehlbauer]

For the tests, I did not write a test for "raise" since it is the default and already being tested by various other tests. If you think it would be beneficial to add another test, then I can write it.

Yes, you are right, this is and was the default. No need to add another test.

[kmuehlbauer]

@pratiman-91 I've thought a bit about the removing of the problematic files for the nested case.

To my understanding we have two corresponding 1D lists (ids and paths1d).

Wouldn't this work to just remember the indices fro the failing files and remove those from the existing lists afterwards? Along the lines:

remove = []
for i, paths in enumerate(paths1d):
   try:
       open(...)
   except:
      if combine="nested":
          remove.append(i)
          
# later in combine nested

if remove:
    for i in sorted(remove, reverse=True):
        del ids[i]
        del paths1d[i]
    

Update, we could even use remove for the checks. No need for another variable

[pratiman-91]

@kmuehlbauer I have thought about this approach before. However, it creates a problem with a 2x2 nested case. Therefore, I have settled for recreating the IDs and paths. You can also check here: 3bfaaee

[kmuehlbauer]

Thanks for the pointer @pratiman-91. Yes, this has a twist.

I see this is a special case for the nested combine. All our good faith doesn't help, for these cases. If the user provides a 2x2, or NxM nested list or whatever and one or more files are missing in one or more of the sublists the whole thing explodes. So I'd rather go for having "raise" in those cases as having the user see ValueError: The supplied objects do not form a hypercube because sub-lists do not have consistent lengths along dimension X.

Maybe others have some opinion here?

[pratiman-91]

@kmuehlbauer @max-sixty Any update on this PR?
Thanks!

-Pratiman