homepage

When using urllib.urlretrieve() to obtain a file
through ftp and the file is not readable the directory
listing of the file is returned. The expected behavior
is that a "permission denied" error is generated.
The current behavior makes it impossible to detect
downloading a file failed.
To reproduce:
from urllib import urlretrieve
fname = "/pub/vim/unstable/testfile"
url = "ftp://ftp.vim.org"
resfile, h = urlretrieve(url + fname)
print "file contents: '%s'" % open(resfile).read()
print "header stuff: ", h
The result:
file contents: '-rw------- 1 506 450 
25 Nov 8 11:50 testfile
'
Using ftplib.FTP() does result in the expected error:
from ftplib import FTP
def list(s):
	print s
f = FTP("ftp.vim.org")
f.login()
f.retrbinary("RETR " + fname, list)
f.quit()
The last line of the resulting exception:
ftplib.error_perm: 550 /pub/vim/unstable/testfile:
Permission denied.

Logged In: YES 
user_id=22785
This is evidently the intended behavior when a url for an
unreadable file is opened, but I think it would make more
sense to raise an exception. To change the behavior so it
raises an exception, simply remove the test for an exception
reason starting with 550 from ftpwrapper.retrfile() in
urllib.py.

Logged In: YES 
user_id=357491
So this was discussed on python-dev (see the thread at http://
mail.python.org/pipermail/python-dev/2004-July/046035.html ) and the 
conclusion was to not change the behavior. Since that handling is the 
only way for directory listings to work when people leave off a trailing 
slash it was deemed not as nice as the current situation.
A note has been added to the "Restrictions" part of the urllib docs to 
state this behavior.