Both recoll and Xapian (the index engine) are written in C++. The document filters are in Python but they only run once per document and tend to be simple and f

Both recoll and Xapian (the index engine) are written in C++. The document filters are in Python but they only run once per document and tend to be simple and fast (easy to add your own too btw). For my use case of 15GB or so PDF files it has been lean and fast. It has a pragmatic Unix tool feel. I am only using the CLI though so can't speak for the GUI. Another Xapian-based tool I use is notmuch (email) and that one is very snappy too.