Using cookies.txt file with Python Requests

Question

I'm trying to access an authenticated site using a cookies.txt file (generated with a Chrome extension) with Python Requests:

import requests, cookielib

cj = cookielib.MozillaCookieJar('cookies.txt')
cj.load()
r = requests.get(url, cookies=cj)

It doesn't throw any error or exception, but yields the login screen, incorrectly. However, I know that my cookie file is valid, because I can successfully retrieve my content using it with wget. Any idea what I'm doing wrong?

Edit:

I'm tracing cookielib.MozillaCookieJar._really_load and can verify that the cookies are correctly parsed (i.e. they have the correct values for the domain, path, secure, etc. tokens). But as the transaction is still resulting in the login form, it seems that wget must be doing something additional (as the exact same cookies.txt file works for it).

Answer 1

MozillaCookieJar inherits from FileCookieJar which has the following docstring in its constructor:

Cookies are NOT loaded from the named file until either the .load() or
.revert() method is called.

You need to call .load() method then.

Also, like Jermaine Xu noted the first line of the file needs to contain either # Netscape HTTP Cookie File or # HTTP Cookie File string. Files generated by the plugin you use do not contain such a string so you have to insert it yourself. I raised appropriate bug at http://code.google.com/p/cookie-txt-export/issues/detail?id=5

EDIT

Session cookies are saved with 0 in the 5th column. If you don't pass ignore_expires=True to load() method all such cookies are discarded when loading from a file.

File session_cookie.txt:

# Netscape HTTP Cookie File
.domain.com TRUE    /   FALSE   0   name    value

Python script:

import cookielib

cj = cookielib.MozillaCookieJar('session_cookie.txt')
cj.load()
print len(cj)

Output: 0

EDIT 2

Although we managed to get cookies into the jar above they are subsequently discarded by cookielib because they still have 0 value in the expires attribute. To prevent this we have to set the expire time to some future time like so:

for cookie in cj:
    # set cookie expire date to 14 days from now
    cookie.expires = time.time() + 14 * 24 * 3600

EDIT 3

I checked both wget and curl and both use 0 expiry time to denote session cookies which means it's the de facto standard. However Python's implementation uses empty string for the same purpose hence the problem raised in the question. I think Python's behavior in this regard should be in line with what wget and curl do and that's why I raised the bug at http://bugs.python.org/issue17164
I'll note that replacing 0s with empty strings in the 5th column of the input file and passing ignore_discard=True to load() is the alternate way of solving the problem (no need to change expiry time in this case).

Answer 2

I finally found a way to make it work (I got the idea by looking at curl's verbose ouput): instead of loading my cookies from a file, I simply created a dict with the required value/name pairs:

cd = {'v1': 'n1', 'v2': 'n2'}
r = requests.get(url, cookies=cd)

and it worked (although it doesn't explain why the previous method didn't). Thanks for all the help, it's really appreciated.

Answer 3

I tried taking into account everything that Piotr Dobrogost had valiantly figured out about MozillaCookieJar but to no avail. I got fed up and just parsed the damn cookies.txt myself and now all is well:

import re
import requests

def parseCookieFile(cookiefile):
    """Parse a cookies.txt file and return a dictionary of key value pairs
    compatible with requests."""

    cookies = {}
    with open (cookiefile, 'r') as fp:
        for line in fp:
            if not re.match(r'^\#', line):
                lineFields = line.strip().split('\t')
                cookies[lineFields[5]] = lineFields[6]
    return cookies

cookies = parseCookieFile('cookies.txt')

import pprint
pprint.pprint(cookies)

r = requests.get('https://example.com', cookies=cookies)