Viktor
Seems I was able to fix several of them, but not all. I think the case with $_GET is still broken, in part because underscores are used internally by the...
Waiting for crawler run to get some PDFs to test against...
PR #196 permits us to export sample data with only PDF files.
PR https://github.com/MarginaliaSearch/MarginaliaSearch/pull/197 Should also count toward this task.
My hunch is we're pulling results by doing a site:domainname.com search, which will include results for all subdomains. If so we should probably add a domain-id limit as well to...
First part was as easy as that, fixed and deployed f19148132afb6a4fc734d18fa12c1c77d5d42294 Second part will require more work. Clearly something is up with the summarizer.
Should be pretty easy to reproduce when there's a known document that causes problems with the summary logic. Though if you find any additional jank, do let me know :)
Fixed as part of PR #99 via dc67c81f9982dfa03deecd1a8e830a6183a65392, will close the issue when that PR goes live.
This ended up being related to encoding issues on the page itself.
Polar.sh is also [open source](https://github.com/polarsource/) and european . Though I'll look into adding it as an alternative.