Improve bencode speed

[ngosang]

https://pypi.org/project/modern-bencode/

decode torrentool_bencode 29.49319815635681 s decode modern-bencode 13.25705099105835 s

Booth libraries are using pure Python code. Maybe you can improve your performance.

def read_torrentool_bencode(torrent_file):
    metadata = torrentool.bencode.Bencode().decode(torrent_file)
    return metadata

def read_modern_bencode(torrent_file):
    metadata = bencode.decode(torrent_file)
    return metadata

torrent_filepath = "aaa.torrent"
with open(torrent_filepath, 'rb') as inp:
    torrent_file = inp.read()

    start = time.time()
    for i in range(0, 100000):
        metadata = read_torrentool_bencode(torrent_file)
    end = time.time()
    print("read_torrentool_bencode", end - start)

    start = time.time()
    for i in range(0, 100000):
        metadata = read_modern_bencode(torrent_file)
    end = time.time()
    print("read_modern_bencode", end - start)

[idlesign]

Thank you.

Maybe you can improve your performance.

It might be so, yet do you have any particular reason to strive for numbers?

[ngosang]

I'm working in my own torrent indexer site, so, I tested several Python libraries. I'm indexing thousands of torrents and serving hundred thousand torrent scrape requests. In this use case performance is important but It's not a common case.

• decode torrentool_bencode 29.49319815635681 s
• decode modern-bencode 13.25705099105835 s
• decode better-bencode 0.98 s

Finally I'm using better-bencode for my project but this is a nice project with more functionality. Maybe I use it for other tasks. This issue is not a priority for me, but can be useful for others. https://github.com/kosqx/better-bencode

[idlesign]

I'm indexing thousands of torrents and serving hundred thousand torrent scrape requests.

I see. Thank you.

decode better-bencode 0.98 s

That's С, not Python.

[idlesign]

Considered closed. Feel free to reopen if required.