Describe the current behaviour
All the document entries in the document table before dynamic batching was erged into main, is going to have empty value in file size column, which means the dynamic batching condition we have on maximum size of one batch to openai's upload and poll api, in this condition our past doc entries will bypass no matter how big they are, and if a user gave big documents to get uploaded to a vector store so a very heavy request will go to openai at once which could cause problems.
Describe the enhancement you'd like
Put a default file size for when we receive a document with empty file size, so that past big files don't add up to too big of a request.
Describe the current behaviour
All the document entries in the document table before dynamic batching was erged into main, is going to have empty value in file size column, which means the dynamic batching condition we have on maximum size of one batch to openai's upload and poll api, in this condition our past doc entries will bypass no matter how big they are, and if a user gave big documents to get uploaded to a vector store so a very heavy request will go to openai at once which could cause problems.
Describe the enhancement you'd like
Put a default file size for when we receive a document with empty file size, so that past big files don't add up to too big of a request.