Repaste content from 0x0.st

[habnabit]

20:49:35 < ChrisWarrick> Could whoever maintains infobob these days put 0x0.st on the list of pastebins to repaste?
                         Firefox wants to download files from that site instead of showing them in the browser
20:51:14 < ChrisWarrick> redsh: 0x0.st uses a dumb text/x-python content type
20:53:23 < ChrisWarrick> re 0x0.st: https://i.imgur.com/yWAwLot.png

[cdunklau]

I tested this a bit. It looks like normally the site sends plain text, but it allows you to specify the content type, e.g.: curl -F'file=@testnullpointer.py' http://0x0.st gave me text/plain, but curl -F'file=@testnullpointer2.py;type=text/x-python' http://0x0.st gave me the specified text/x-python. Now, that was with super simple files, just imports. I made another one:

import sqlite3

def main():
    with sqlite3.connect(':memory:') as conn:
        conn.execute('CREATE TABLE foo (name INTEGER PRIMARY KEY)')


if __name__ == '__main__':
    main()

...and still got plaintext, but adding #!/usr/bin/env python to the top make the site autodetect and gave me that x-python thing again.

This behavior and the possibility of arbitrary file types makes adding full support to infobob nontrivial, but I could imagine some kind of simplistic detection technique for a limited set of "file types" (the extension in the URL?). One idea could be:

1. If the 0x0.st url ends in .py, download it, check the content type. Probably should limit downloads to something reasonable to avoid needless bandwidth (and memory) consumption, say, 16KiB? I think bpaste's limit is around there anyway.
2. If the content type is text/plain, ignore the URL.
3. Else assume it's python, and repaste.

However, based on my logs from this year to date, this site was only used ~27 times, 12 of which were .py URLs. That puts it rather low on my priorities. I'd certainly be willing to merge this feature though. @Kwpolska are you feeling up for it?

[Kwpolska]

@cdunklau I gave it a shot, PR #63.