Currently chunks are just bare dictionary objects returned from Library. this makes it hard to refactor how they're represented and what properties they have.
Related to #15 . Helpful for #56, #33, and others.
[x] Update Library.chunks() to return an array of Chunk, not a list of (chunk_id, chunk) (now that chunk.id holds the id)
[x] Have Chunk be the thing that converts the embedding from base 64 to fully expanded
[x] Have Library have an internal dict of cached id -> chunk objects
[x] Add Chunk.remove() which calls chunk.library.remove_chunk(self)
[x] chunk.str should return chunk.text
[x] Cache chunk.id() if it uses canonical id
[x] Have a Chunk.validate() that is called on chunk creation, and move the Chunk validation from Library.validate() to that (when Library is created a Chunk wrapper should be created around each)
jkomoros
commented
1 year ago
Currently chunks are just bare dictionary objects returned from Library. this makes it hard to refactor how they're represented and what properties they have.
Related to #15 . Helpful for #56, #33, and others.