As already discussed on GitHub (https://github.com/hibernate/hibernate-search/pull/1634)
While updating to a more recent version of Tika (1.17), we cannot rely on the default AutoDetectParser anymore, as we need to define a custom Tika instance with a special tika-config.xml to drop a specific parser (GrobIdParser, https://wiki.apache.org/tika/GrobidJournalParser) which, at least after my lib update, seems now the default for indexing PDFs. This is effectively suggested in
In my case, I had to provide a custom tika-config.xml:
To reflect this customization in the TikaBridge as well, some additional handle for this is required as the AutoDetectParser is hard-wired in TikaBridge.