Sorry for the late reply. I don't know how this works in Solr, but having to strip out the special cases before feeding to FTS5 would require a custom function at compile time right? Otherwise I fear there would be a lot of waste with data duplication by having to store the "sanitized" data into another column.