In a scope of this spike we need to review database structure, remove unused columns, possibly reorganize tables.
Not full mod-kb-ebsco-java database structure:
Unused columns and tables:
Providers, Packages, Titles, Resources.
All this objects we retrieve from EBSCO HoldingsIQ. And module did not store them until we add tag to one of them.
- Module storing records if they have tags, if tags list is empty, module will delete it.
- Module retrieve records by tags. And then request them by id from HoldingsIQ.
It looks like this 4 tables are needed just for tags. And we using only id column. While other columns (like name, content-type) are inactive.
I suggest to remove providers, packages, titles, resources tables. Because we already have record_id and record_type columns in tags table.
One addition is to include credentials_id column to tags
We use a lot of joins to retrieve records by tags. But now we can retrieve records from one table.
- Write databaseChangeLog script to edit database structure
- Delete repositories related removed tables PackageRepository, ResourceRepository, TitleRepository, ProviderRepository
- Implement logic in TagsRepository to retrieve records from tags table by TagFilter
- Unit tests refactoring
- Remove unused data, simplify database structure
- Get rid of code duplications, so searching by tag logic will be in the one class. Instead of 4 same classes
- Will stop using SQL Joiners to improve module response time when searching by tags.