reduce torrent filename length

shorten.py

from bcoding import bencode, bdecode
import sys
import time
import os
import pathlib

remapped = {}
rename_list = []


def shorten_filename(name, length):
    left = 0
    right = len(name)
    middle = 0
    while right != left and right - 1 != left:
        middle = int((left + right)/2)
        if len(name[:middle].encode('utf8')) > length:
            right = middle
        else:
            left = middle
        # print(left, right, middle, len(name[:middle].encode('utf8')), length)
    return name[:middle-1]

bt = None
with open(sys.argv[1], 'rb') as fp:
    bt = bdecode(fp)
    for file in bt['info']['files']:
        changed = False
        original = file['path'][:]
        for index, part in enumerate(file['path']):
            t = tuple(file['path'][:index+1])
            if t in remapped:
                file['path'][index] = remapped[t]
                continue
        
            if len(part.encode('utf-8')) > 255:
                name, ext = os.path.splitext(part)
                capped = shorten_filename(name, 254 - len(ext.encode('utf8')))
                new_name = capped + ext
                file['path'][index] = new_name
                new_t = tuple(file['path'][:index] + [new_name])
                remapped[new_t] = new_name
                changed = True
        
        if changed:
            rename_list.append(("/".join(original), "/".join(file['path'])))

for l, r in rename_list:
    print(l, '->', r)

if rename_list:
    original = pathlib.Path(sys.argv[1])
    original.rename(original.with_name(original.name + '.bak'))
    with open(sys.argv[1], 'wb') as fp:
        bencode(bt, fp)

Note: infohash is calculated based in info's bencode, changing name in info dictionary resulting in tracker reject modified torrent.