filter: --group-by fails to error properly when used with date-only values

[victorlin]

While writing tests for #794, I noticed a couple edge case-y bugs.

Setup

Create a simple metadata file without a date column:

echo 'strain\tcountry
1\tA
2\tB' > metadata.tsv

Group by year

augur filter \
    --metadata metadata.tsv \
    --group-by year \
    --subsample-max-sequences 10 \
    --output-strains out.txt

Output:

WARNING: The specified group-by categories (['year']) were not found. No sequences-per-group sampling will be done. Note that using 'year' or 'year month' requires a column called 'date'.
Sampling at 10 per group.
Traceback (most recent call last):
  File "/opt/homebrew/Caskroom/miniconda/base/envs/nextstrain-native/bin/augur", line 33, in <module>
    sys.exit(load_entry_point('nextstrain-augur', 'console_scripts', 'augur')())
  File "/Users/vlin/repos/augur/augur/__main__.py", line 10, in main
    return augur.run( argv[1:] )
  File "/Users/vlin/repos/augur/augur/__init__.py", line 75, in run
    return args.__command__.run(args)
  File "/Users/vlin/repos/augur/augur/filter.py", line 1537, in run
    queues_by_group[group].add(
KeyError: ('_dummy',)

This should fail earlier instead of trying to use _dummy.

Group by year and month

augur filter \
    --metadata metadata.tsv \
    --group-by year month \
    --subsample-max-sequences 10 \
    --output-strains out.txt

Output:

WARNING: A 'date' column could not be found to group-by year or month.
Filtering by group may behave differently than expected!
Sampling at 10 per group.
WARNING: A 'date' column could not be found to group-by year or month.
Filtering by group may behave differently than expected!
0 strains were dropped during filtering
	0 of these were dropped because of subsampling criteria
2 strains passed all filters

This should fail instead of passing.

[huddlej]

Thank you for calling this out, @victorlin! This might be a subset of #754, unless you think there's enough difference to keep this issue open. I'd love to see better error handling in augur filter (and generally in augur). Some of the current implementation decisions date back to the first version of Augur and could definitely be revisited.

[victorlin]

Didn't mean to close this 🤦

[victorlin]

@huddlej

This might be a subset of #754, unless you think there's enough difference to keep this issue open.

When I opened this, the KeyError: ('_dummy',) was a different behavior so keeping this issue separate made sense. However, the behavior is consistent now after merging #794. Closing this issue.