Create high-level list of unit tests

[agstephens]

Basic unit tests

These tests confirm things are working:

• 1. item search returns correct count
• 2. asset search returns correct count
• 3. asset search within one item returns correct content (number of assets, supports time filtering etc) - using .get_asset(id), .assets, .get_assets().
• 4. item search on instance_id returns 1 item, matching instance_id (e.g.: CMIP6.HighResMIP.MOHC.HadGEM3-GC31-HH.highres-future.r1i1p1f1.day.tas.gn.v20191105)
• 5. assert that there are no duplicate asset records in an item (using instance_id above)
• 6. item search on various facets matches those facets in the results
• 7. Unit test: temporal search returns correct content #8

Comprehensive unit tests

These tests explore the breadth of options in the search client:

• 1. POST and GET requests work the same
• 2. not testable until multiple collections indexed - get a collection object, then specify it in the search with client.search(collections=[my_collection])
• 3. Unit test: get two item _objects_, then specify them in the search with client.asset_search(items=[item1, item2]) #9
• 4. limit=, max_items= to specify the number of items to return in a response, limit is items per page and max_items is total return. (default: limit=100, max_items=100) #10
• 5. filter=, filter_lang= to include a search query in the filter parameter. (default GET search: filter_lang=cql-text, filter=None) #13 no need to test filter_lang - only 1 lang implemented so far
• 6. complex searches work, such as #14
  • Search for all datasets that have variable ta OR hus, excluding those with acitivity_id of AerChemMIP:
    • client.search(filter="activity_id<>'AerChemMIP' AND (variable='hus' or variable='ta')")
• 7. to filter by a string against values in the properties. (default: q=None) #11
  • 8. test various search fields/terms across facets, including wildcards with "*"
• 9. Unit test: map between ESGF terms and STAC terms using "doctype" #7
• 10. not testable until multiple collections indexed - walking through levels works, e.g.: client.get_collection(collection_id).get_item(item_id).get_asset(asset_id)
• 11. when you search with latest=True (at the item-level): the response should only give back the most recent version of any given dataset. latest=True or latest='true' raises an error and does not work as expected
• 12. search by version works #12
• 13. bbox and intersects
• bbox=[West,South,East,North] to filter within a geo-spatial box. (default: bbox=None)
• intersects= to filter by any geo-json type. (default: intersects=None)
• 14. allow filtering of properties returned in results, e.g. #15
  • 15. check that fields and source can be used interchangeably
• 17. searching on cmip6 vocabulary matches responses when searching using general vocabulary (refer to: Fully describe the plan for searching and indexing with name-spaced facets (vocabs) search-futures#107)
• 18. make sure pagination works for both item and asset search

[agstephens]

@rhysrevans3 you might want to use pytest.mark.parametrize like here:

https://github.com/cedadev/midas-extract/blob/master/tests/test_midas_extract.py#L98-L106