Well, nodejs itself is not very suitable for this, although I could be wrong.
I would advise looking towards the server browser.
Something like Phantom.js
But there are higher-level (more convenient to work with pages and in general) phantom-based wrappers, such as Casper .
Here is something very simple and recently released by Nightmarejs

In a very primitive case, authorization for the client works like this:
- the user submits the form data in a POST request. That is, the data is written to the request body, not to the headers.
- if we talk about PHP, then when the session_start function is called (or automatically if set in php.ini), a session file is created (again, it can be stored in memory in Reddis, as configured) and the user is assigned a cookie with a session ID. This same cookie is set through the response headers:
Then, with each request to the server, you will also need to send these cookies in the headers, otherwise the server will think that you are a new dude and will create a new session for you.
In the session, on the server, everything that is connected with authorization is usually stored. That is, in most cases, you just need to send the cookies that you set from the server and resolve this matter.
But alas, this is only the most common method of authorization, as far as I remember, in the same IPB, sessions were usually additionally tied to your IP and a bunch of other things, such as separate tokens in the URI, but I honestly don’t remember anymore, I haven’t worked with these things for a long time.
In your case, the simplest solution would be to arm yourself with phantomjs and just automate user actions. All these things like saving and handling cookies are supposed to be done for you. You just need to write down the sequence of actions. It will also be much easier to parse.

Have you looked at the finished modules? roboto , crawler , simplecrawler ?