Update for PyMongo 4 and add GitHub Actions

[blink1073]

Description of changes

• Adds GitHub Actions testing
• Adds Python 3.10 support - Fixes Python 3.10 compatibility #841
• Adds pymongo 4.0 support - Fixes mlaunch: Test with PyMongo 4.0 #844
  • Adds handling of directConnection
  • Fixed use of pymongo.database.Database.authenticate
  • Fixed use of pymongo.database.Database.add_user
  • Fixed use of pymongo.collection.Collection.insert
  • Fixed use of pymongo.collection.Collection.update
  • Fixed use of pymongo.collection.Collection.ensure_index
  • Fixed use of pymongo.collection.Collection.count
  • Fixed use of pymongo.collection.Collection.group
• Adds pytest support (needed for GitHub Actions support) - Fixes Replace Nose with an actively maintained unit testing library #842

Testing

• GitHub action on my fork
• Locally running tox on macOS

O/S testing:

O/S	Version(s)
Linux	Ubuntu 20.04.3
macOS	10.15.7

[blink1073]

Converting to draft while I continue to work through the pymongo 4.0 changelog

[stennie]

Epic! Thanks for the PR @blink1073 !

[stennie]

Nice monkeypatch. Ideally we should be unit testing output of run() rather than grubbing from stdout, but we can fix that later.

[blink1073]

I went through all of the deprecated functions, and the build is passing, taking this out of draft

[ShaneHarvey]

This fails because the 'shard' field doesn't exist:

>>> chunk_group = client['config']['chunks'].aggregate([{ '$match': { 'ns': 'test.test2' }},{ '$group': { '_id': '$shard','nChunks': { '$sum': 1 }}}])
>>> list(chunk_group)
[{'_id': 'demo-set-0', 'nChunks': 1}]
>>> chunk_group = client['config']['chunks'].aggregate([{ '$match': { 'ns': 'test.test2' }},{ '$group': { '_id': '$shard','nChunks': { '$sum': 1 }}}])
>>> print(', '.join(["%s: %i" % (ch['shard'], ch['nChunks']) for ch in chunk_group]))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 1, in <listcomp>
KeyError: 'shard'

To fix this we need to add the shard field which is a little awkward in the $group stage. AFAICT you need to use the $first accumulator:

>>> chunk_group = client['config']['chunks'].aggregate([{ '$match': { 'ns': 'test.test2' }},{ '$group': { '_id': '$shard','nChunks': { '$sum': 1 }, 'shard': {'$first': '$shard'}}}])
>>> print(', '.join(["%s: %i" % (ch['shard'], ch['nChunks']) for ch in chunk_group]))                                                                      
demo-set-0: 1

Also this code will not work at all on 5.0 because they removed the "ns" field from the chunks collection. Compare https://docs.mongodb.com/v5.0/reference/config-database/#mongodb-data-config.chunks with https://docs.mongodb.com/v4.4/reference/config-database/#mongodb-data-config.chunks . The config database is internal so can change from release to release. This particular issue should be handled in a new ticket.

[ShaneHarvey]

Sorry, looking at this again there's no need for the 'shard': {'$first': '$shard'}. We're already grouping on the shard field so we can change the list comprehension to use the _id field:

>>> chunk_group = client['config']['chunks'].aggregate([{ '$match': { 'ns': 'test.test2' }},{ '$group': { '_id': '$shard','nChunks': { '$sum': 1 }}}])
>>> print(', '.join(["%s: %i" % (ch['_id'], ch['nChunks']) for ch in chunk_group]))                                                                      
demo-set-0: 1

[ShaneHarvey]

The pymongo related changes LGTM!