I have no solution really. In my experience html is a bit too variable. I can't find anything DBI-ey (also looked in https://pgxn.org/ - no luck ). Of course, to get database-access you could slurp either html (via curl) or cleaned-up text (via links -dump) into a table but they'd just be 'raw' lines that you'd still have to select the correct table rows from. Still, for well recognizable/greppable rows it might work. And anyway, it is a reminder that postgresql's COPY knows how to read input from another program's STDOUT.
create table temp_slurps (line text);
copy temp_slurps ( line ) from program 'links -dump -width 512 ${url}'
+;
select * from temp_slurps ; -- where ...
As they say, YMMV. I'm sure if you write a postgres extension (for pgxn.org) to extract-'read' html-tables from source it will be popular ;)