allisfine
Just one thing: I see you are splitting by |, and some torrents (very few, but some) have | in their name (I didn't bother with escaping that).
But it would still be more proof of concept than really anything useful - the comments and descriptions ARE important.
edit: More I am thinking about it, the less useful it sounds.
First, the information about seeders vary constantly, especially with the new torrents.
Also, it STILL depends on single point of failure - the Pirate Bay itself. If TPB will be down for any reason, I will have no place to scrape this from and it will all fall apart anyway.
Plus, I think Pirate Bay itself should make dumps like this. It would probably be much better for their database anyway :)
The archive is static, the TPB database is dynamic :)
but not 100%
I went trough the description pages like http://thepiratebay.se/torrent/$i by increasing the $i and saving the magnet if pirate bay didn't return 404 error. I went trough the pages as unlogged user, though, so that might be the reason why I got only 1.5m torrents.
I didn't know pirate bay has hidden porn torrents; there is TONS of porn in the scrape already.
The script is in perl, I will post it to pastebin in a moment.
edit: allright, the script itself is here http://pastebin.com/8RXXthXB
as you can see, it's not very complicated.