Support for TM1SessionID or cookiejar in RESTService

[ajmyers]

It would be useful if TM1py supported initializing a RESTService with an additional TM1SessionId or session cookiejar parameter to bypass the authentication process on the TM1 server. An LDAP authentication request can take 10-20s for us (or can be delayed if there are waits on the TM1 server), so caching the session is a necessity for us.

[MariusWirtz]

Yes. That would be a nice feature.

I think the most intuitive way to do it is through an optional string argument (session_id) in the TM1Service / RESTService Constructor.
Shouldn't be more than a few extra lines of code in the RESTService constructor.

How come you have a sessionid at hand before initializing TM1py?

[ajmyers]

Currently, i'm grabbing the session cookie from RESTService._s.

For the time being, I'm doing an ugly patch of RESTService._get_cookies to add in basic persistent session support via exporting/importing to/from a temp file, however a supported mechanism to export the session and reinitialize TM1Service or RESTService with that session would be ideal.

def _get_cookies(self):
   """ perform a simple GET request (Ask for the TM1 Version) to start a session
   """
    url = '{}/api/v1/Configuration/ProductVersion/$value'.format(self._base_url)

    auth_string = self._headers['Authorization'].split(" ")[1]
    auth = b64decode(auth_string).decode('ascii').split(":")
    if 'CAMNamespace' in self._headers['Authorization']:
        session_name = '{}_{}_{}'.format(self._base_url, auth[0], auth[2])
    else:
        session_name = '{}_{}'.format(self._base_url, auth[0])

    cookie_path = tempfile.gettempdir()
    cookie_file = os.path.join(cookie_path, urlsafe_b64encode(session_name.encode()).decode() + ".tm1_session")

    try:
        with open(cookie_file, 'rb') as f:
            cookies = pickle.load(f)
            requests.utils.add_dict_to_cookiejar(self._s.cookies, cookies)
            h = self._headers.copy()
            h.pop('Authorization')
            response = self._s.get(url=url, headers=h, data='', verify=self._verify)
            self.verify_response(response)
            self._version = response.text
            return
    except TM1pyException as e:
        print('Bad cookie, generating new session')
    except Exception as e:
        print('No cookie or some other error, generating new session')
        pass

    response = self._s.get(url=url, headers=self._headers, data='', verify=self._verify)
    self.verify_response(response)
    self._version = response.text

    with open(cookie_file, 'wb') as f:
        pickle.dump(requests.utils.dict_from_cookiejar(self._s.cookies), f)

[MariusWirtz]

Thanks for the sample code! I like the use of pickle.

I am thinking about serializing the full TM1Service instance into a pickle file.
I would like to do it by introducing two new functions to the TM1Service:

• one that serializes the TM1Service instance (including the session cookie) into a pickle file.
• and another one that de-serializes the TM1Service from a pickle file. I think we could build the second one as a classmethod (independent constructor) of the TM1Service.
  So codewise it could look like:

tm1 = TM1Service.from_file(path_to_tm1service_pickle)
print(tm1.cubes.get("plan_BudgetPlan"))

I think it would serve the purpose and it should be easy and intuitive.
What do you think?

[ajmyers]

Yes, that sounds perfect!

[MariusWirtz]

Hi,

I pushed the new release to Github and PyPI.
Now we can dump the tm1service instance to a file and reuse the session later.

import time
from TM1py.Services import TM1Service

# Connect to TM1
tm1 = TM1Service(address='localhost', port=12354, user='admin', password='apple', ssl=True)
print(tm1.server.get_server_name())
# Save TM1Service instance to file
tm1.save_to_file('tm1_connection')

time.sleep(7)

# Restore TM1Service instance from file
tm1 = TM1Service.restore_from_file('tm1_connection')
print(tm1.server.get_server_name())

# Logout
tm1.logout()

[ajmyers]

@MariusWirtz This is awesome, thank you!