May 12
So, I lied.
4scrape is alive again. Completely rewritten from scratch in Go, featuring even more stupid bullshit on the frontend. And it’s using a real fulltext indexer this time around. In it for the long haul this time around.
Tagged with: 4scrape, cake
15 comments
15 Comments so far
Leave a comment

Oh, that was you, then! I posted on warosu saying that I liked it, but I got accused of being you. :<
Hope you're ready for 4chan's HTML changes that are supposedly going to be rolled out today, hah.
I’m not ready for the changes, actually. And the scraper is running in the same process as the HTTP server (I’m going to implement real-time scrape viewing and shit with websockets) so we’ll see what happens!
PROBABLY SIGSEGV.
WELP.
Also, sorry for beating you to the punch and totally ninja’ing the name Asagi!
what the fuck. where did this come from.
Also, do you write any Haskell nowadays?
> where did this come from
Pulled it out of my ass a couple weekends ago. Kinda missed having it around.
> Haskell
Not really, no. Haskell is fun and all, but it forces you to spend a lot of time up-front coming up with a good design for the program. I found myself fighting with the compiler more often than not. My language du jour is Go these days.
Shrug.
Ok. I was reading some old posts about 4scrape/haskell and I was thinking about learning it.
Will there be a larger write-up of the setup? How are you hosting it, what are the plan for the future, any pitfalls when using Go, and so on?
Haskell isn’t bad; I’d say it’s a good language to learn, but not to necessarily use in production environments.
I’ll probably do a write-up at some point; it’s hosted on a dedicated machine leased from 100TB.com. With respect to Go, the only nasty snags I’ve encountered have been semi-shoddy library support:
Other than that, everything has been pretty straightforward. The code’s a fucking nightmare right now (just read main.js — the server is roughly as bad), but it’s mostly just a matter of refactoring. I’m more interested in adding back features than making the code look pretty.
Ok cool.
I’ve noticed some problem with the filter menu, mainly with dropdowns that don’t work (can’t select stuff || after selection it won’t register clicks) but other than that it’s very nice.
The filters aren’t applied retroactively — they’re used whenever an API call is made, which means you (currently) have to either scroll past what’s on the screen or click the ‘Home’ button (or do another search). I’m going to add an extra button that wipes what you’re looking and and re-performs the API call sometime this week.
Oh btw, what are your thoughts on the raspberry pi?
They can’t take my money fast enough. I want to strap an IMU to it, write a shader-based integrator and build a quadcopter frame around the things. I have pretty weak plans to do the same to my pandaboard, but the design may be prohibitively large.
I dig the older style more, checkout nik.bot.nu the way he adapted the original code for 4scrape just looks simply amazing.
Nice job man.
4scrape is amazing, you rock.
I suggest a feature to filter out /mlp/ content.