Looking for testers: index_server revival (Spotlight-like search for Haiku)

some of you might remember index_server from Clemens Zeidler’s GSoC project back in 2010 — a background service that watches your files and keeps them indexed, kind of like Spotlight on macOS. It got mostly abandoned after that, but I worked on it six yers ago nearly got it to a working stage (except of recursivly scanning the own index :-D)…
I’ve been picking it back up these last weeks with the help of ai… and wanted to share where it’s at.

What it does right now:

  • Watches your volumes and analyzes files as they change

  • Full-text search over file content (not just filenames) via CLucene, with a small standalone search app

  • Thumbnails for images (including our own HVIF icons) that show up directly in Tracker’s icon view

  • EXIF data for photos, audio tags, media codec/duration info — all written as normal BFS attributes, so they show up in Tracker too

  • Mail body content is searchable as well

  • A settings preflet for exclude paths and enabling/disabling individual analysers

I spent the last weeks and some money on ai… chasing down a bunch of real bugs (a few actual crashes, some silent indexing failures, non-ASCII filenames not working, thumbnails not showing at certain icon sizes, that kind of thing), so I think it’s in decent enough shape now for a first proper round of testing outside my own VM and my own real installation.

This is a debug build — full debug symbols, and it logs everything to ~/config/var/log/index_server/index_server.log so if something breaks I actually have a chance of figuring out why. If you hit a crash, Haiku’s debug_server will offer to save a report on your Desktop — please grab that too if you can.

Would really appreciate anyone willing to install it and just… use your Haiku like normal for a while and see what happens. Especially interested in:

  • does it survive being left running for a long time

  • how it handles a first full scan of a “real” (i.e. messy, years-old) home directory

  • anything that crashes or seems to silently not index stuff it should

Download + install instructions + full changelog: https://github.com/Paradoxianer/haiku/releases/tag/index-server-debug-1

(fair warning, first backlog scan of a big volume can take a while — there’s a progress notification so you can see it’s actually working)

Thank you .. looking forward to your opinion…
One of my most loved examples … Just to give you a glimpse what it would be able to Install tesseract orc translator and it should be able to even search for text in images … :slight_smile:

4 Likes

Great stuff!

Does the index server have an API that apps can use to query a specific file type (I am interested in email files) and get back the results for processing in the app?

At the moment it comes with a index searcher but this app work this way that it sends a BMessage with a "query" string (plus optional "maxResults") to the index_server and get back matching entry_refs + scores. so no “filter” if the file is a e-mail type…

@Paradoxon

Thank you for the reply! Keep up the good work!