[issue34951] cookielib/cookiejar cookies' Expires date parse fails with long month names

2018-10-10 Thread Alberto Moral 


New submission from Alberto Moral :

http.cookiejar (cookielib, for python2.*) does not parse some cookies' Expires 
date.

For  example: "Friday, 1-August-1997 00:00:00 GMT" does not work (while: "Fri, 
01 Aug 1997 00:00:00 GMT" works fine)

This is basically due to long names of months (it is compared with 
MONTHS_LOWER: list of 3-letter months). So, I propose a small change in the 
definition of LOOSE_HTTP_DATE_RE (see fifth line):

LOOSE_HTTP_DATE_RE = re.compile(
r"""^
(\d\d?)# day
   (?:\s+|[-\/])
(\w{3})\w* # month (3 first letters only)
...

Instead of:
LOOSE_HTTP_DATE_RE = re.compile(
r"""^
(\d\d?)# day
   (?:\s+|[-\/])
(\w+)  # month
...

I've tested only http.cookiejar (python 3.6), but I suposse the same change 
will work on cookielib

Thanks in advance

--
components: Library (Lib)
messages: 327461
nosy: alb_moral
priority: normal
severity: normal
status: open
title: cookielib/cookiejar cookies' Expires date parse fails with long month 
names
type: behavior
versions: Python 2.7, Python 3.6

___
Python tracker 
<https://bugs.python.org/issue34951>
___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue34951] cookielib/cookiejar cookies' Expires date parse fails with long month names

2018-10-10 Thread Alberto Moral 


Alberto Moral  added the comment:

Thanks for your answer. I have not found any RFCs with full month names either. 
I'm afraid I'm not an expert here.

But the case is that I get them in my work. Here is an example of response 
header:

  HTTP/1.1 200 OK
  Server: Oracle-iPlanet-Web-Server/7.0
  Date: Tue, 10 Oct 2018 14:29:44 GMT
  Version-auth-credencial: v.3.0.1 Iplanet - Sun Solaris - Contexto Multiple
  Set-cookie: JSESSIONIDE=Del; expires=Friday, 1-August-1997 00:00:00 GMT; 
domain=...

I do not know if it's an old date format (?)... or if it is a quite rare case...

I have created some previous bash scripts using wget and they work fine, but I 
have had problems with python3 (and requests module) till I realized this 
issue. And it was not very easy: I am very new with python :( 


That's the reason of my proposal. It's just to be coherent: if we compare 3 
letters of a month with MONTHS_LOWER, let's use just 3 (first) letters.

Perhaps modifying LOOSE_HTTP_DATE_RE is not a good idea. Another option could 
be to truncate the month variable (mon).

It could be done inside the _str2time funtion, for example:

def _str2time(day, mon, yr, hr, min, sec, tz):
mon = mon[:3]  # assure 3 letters
yr = int(yr)

Anyway, I'll try to find why those long month names appear.

Thank you

--

___
Python tracker 
<https://bugs.python.org/issue34951>
___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue34951] cookielib/cookiejar cookies' Expires date parse fails with long month names

2018-10-10 Thread Alberto Moral 


Alberto Moral  added the comment:

Yes, I was thinking that it could be a matter of configuration of the server 
(?).

By the way, and just for fun, I've just realized that truncating mon at the 
begining of the _str2time funtion is a very bad idea because mon could also be 
an int.

A better place is when looking the MONTHS_LOWER array index (and possible 
exception is handle):
try:
mon = MONTHS_LOWER.index(mon[:3].lower())+1

(perhaps in 2 sentences for clarity)

OK, waiting for experts' comments.

I'm really enjoying Python.

--

___
Python tracker 
<https://bugs.python.org/issue34951>
___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com