Open
Description
As implemented in #79 the resource_source_as_path
now always copies the source file into a (potentially temporary) workdir, even if the file is a local file and already has a path. This means that when one runs docling *.pdf
with a lot of pdfs in the current working directory, docling spends a lot of time copying them over to a temporary directory. In one of my use cases the /var/tmp
has per-user quotas and docling processing with a lot of PDFs has reached that quota and errored-out. I believe the default should be to not do a copy, or at least, there should be a way to request not doing a copy (via the CLI and also the library).
Metadata
Metadata
Assignees
Labels
No labels