VW3 Import Creates Duplicate Categories & Attachments
When importing from VW3, it's possible that:
- User used ATTACH in a post, when the ATTACH points to a valid wiki attachment. If the attachment has not been created yet, finding the ATTACH creates it. But then the wiki attachment importer creates it later. So now there are two copies.
- User used CATEGORY in a page. If the category has not been created yet, finding the CATEGORY creates it. But then the wiki page importer creates it later. So now there are two copies of the category, each with some members that should only be in one or the other.
To fix, links should NOT be processed on the first pass importing wiki pages. Instead, all pages and all edits should be imported first. Then a new step should be added to reprocess the current revision on every page to find categories and attachments.
In particular the CATEGORY issue leads to data loss, because the URL conflict must be resolved, and in doing so, some or all of the current category associations for the conflict would be lost.