DuckDB
does provide a waut to delete nodes based on doc_id. vector_store.delete(documents[0].doc_id)
to delete a document loaded in current session. i don't know the doc_IDs of documents in a persisting database. I guess correct question is, how to know doc_IDs of documents in a persisting DuckDB database file so that I can delete them.From where did you get this documents object?
I don't have doc object as I am loading a existing vector store from a persisted directory. That's where i am stuck at. How to get doc_ID for a document that i want to delete from a persisted DuckDB vector store.You can do simple query and the response contains retrieved nodes which contains node ID then you can try deleting them.
how to get doc_IDs from retrieved node objects? Ain't i need doc_IDs to delete?