If the attachment doesn't get flagged as containing potential PII there are cases where we're comfortable in republishing the attachment automatically, exceptions are:
If XLS with pivot cache - the conversion to XLSX means the pivot cache is lost
With named ranges and external links present - the data cached isn't being scanned for PII
With a data model present - we can't, at present, scan the data inside this data model.
If the attachment doesn't get flagged as containing potential PII there are cases where we're comfortable in republishing the attachment automatically, exceptions are:
Fixes https://github.com/mysociety/whatdotheyknow-private/issues/312