https://git.savannah.gnu.org/cgit/gawk.git/tree/doc/gawktexi.in Current approach using html parsing is very brittle.