MITLibraries / timdex-index-manager / 19441567561

push

github

Stub CLI command for updating existing docs with embeddings (#370)

Why these changes are being introduced:
* TIM requires a CLI command to update existing docs with
embeddings. Currently, TIM uses the OpenSearch
'index' action to create/update docs using *full* records,
which performs a full replacement of existing docs. To update docs
using *partial* records (e.g., only a subset of TIMDEX fields
[i.e., embeddings]), the OpenSearch 'update' action must be used.

How this addresses that need:
* Add 'bulk_update_embeddings' CLI command
* Create BulkOperationError exception
* Update tim.helpers.generate_bulk_actions to format inputs for
OpenSearch 'update' action

Side effects of this change:
* TIMDEX Pipeline Lambdas will require updates to generate
the 'bulk-update-embeddings' CLI command.

Relevant ticket(s):
* https://mitlibraries.atlassian.net/browse/USE-122

17 of 57 new or added lines in 4 files covered. (29.82%)

430 of 481 relevant lines covered (89.4%)

0.89 hits per line