Changes in this release:- Strip non-ascii characters from the pdfreflow xml which caused it to be invalid
- Support the ^ character being part of the ISBN number
- Attempt to minimise any memory leak issues caused by this plugin itself
Note that as per my post above there are still memory leak issues with some of the "conversions" that get run in the background. I've made sure that the plugin releases all file handles and resources that it creates asap so anything else is in the Calibre code. To try to work around that will require greater changes to this plugin than I want to make at this point but I will likely revisit it in future once I finish another new plugin that will work in a similar way.