Someone noticed above that "gzip" should be set to False in line 56. It looks like that the "gzip=True" parameter was supposed to decompress the file before saving it to /tmp, but it didn't. So the whole installation script can be fixed by manually decompressing, and you don't need to download it yourself. Just make sure getedict.py looks like this (the 5 lines starting with "if sknetio" are the changed/added lines):
with SkProfiler("fetch"):
# gzip=True to automatically extract gzip
# flush=false to use more memory to reduce disk access
if sknetio.getfile(url, path + '.gz', flush=False, gzip=False):
import gzip, shutil
with gzip.open(path + '.gz') as f_in, open(path, 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
skfileio.removefile(path + '.gz')
ok = skfileio.filesize(path) > minsize
if not ok and os.path.exists(path):
skfileio.removefile(path)
A very similar problem exists for KanjDic, and the same exact code can fix it. In getkanjidic.py:
with SkProfiler("fetch"):
# gzip=True to automatically extract gzip
# flush=false to use more memory to reduce disk access
if sknetio.getfile(url, path + '.gz', flush=False, gzip=False):
import gzip, shutil
with gzip.open(path + '.gz') as f_in, open(path, 'wb') as f_out:
shutil.copyfileobj(f_in, f_out)
skfileio.removefile(path + '.gz')
ok = skfileio.filesize(path) > minsize
if ok:
os.renames(path, targetpath)
I could have fixed sknetio.py instead to actually work with gzip=True but oh well. I'm pretty sure this is it anyway.