Add support to read and assign ID for Document object directly from a CSV column in CSVLoader

[muhammadyaseen]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

In the CSVLoader, we currently have no way to specify an id_column in the constructor here.
I would like to add this functionality to the class.

Motivation

In many applications, there is a natural document ID already available in the CSV data e.g. experiment_id, conversation_thread, internal_document_identifier etc. It would be nice to have a consistent ID being used in the vector store as well as in the system where the data originates from.

Proposal (If applicable)

Add an optional argument id_column: str | None to the constructor and if it is set, read and assign that ID to the created Document object. Any error handling and checks that this requires will also be added e.g. if the column is not present or if there are duplicates etc.

I will be happy to contribute to this.