Extend model object with request context

[basimhennawi]

Problem:

There are many cases where you want to pass some parameters and access them in the model methods but you're only limited to the method signature params usually some attribute inside the request object. After going through some similar-issues (#254, #336) and PRs (#320) found out there are a lot of usecases needed for different people implementing this lib.

Example:

In saveToken method I need to pass and save consentId attribute in the token table from the request body, as follow:

const saveToken = (token, client, user) => {
    const { consentId } = this.request.body;
    ...
}

Suggested solution (Quick-win):

Extend model object with request context in the handlers (authorize, authenticate, token) so that it can be accessible in any of the model methods, also it won't break the current tests.

Note:

Right now, I forked the lib and did this minimal change as a quick-win but it'd be better if this issue is fixed on the lib so I still can refer to the original lib no my forked one.

Any feedback or change requests are welcome! Thanks!

[mjsalinger]

Looks good, just two things:

a) Pleasechange this PR to a merge request to the dev branch instead of `master.
b) Please add some tests with regards to setting and accessing the request context from model methods.

[mjsalinger]

Great, if you add some tests I think we can merge this.

[basimhennawi]

Unit and integration tests are added! Thanks @mjsalinger for reviewing!

[mjsalinger]

LGTM

[thomseddon]

I don't think this works as expected as the model functions can be executed asynchronously, the request as set here could be overwritten during model execution.

For example:

const model = {
  request: false,
  saveToken:  function (num) {
    console.log(`saveToken for request ${num} start - model.request = ${model.request}`)
    setTimeout(() => {
      console.log(`saveToken for request ${num} end - model.request = ${model.request}`)
    }, 2000);
  }
}

// 1st request received 1 second after start
setTimeout(() => {
  model.request = 1
  model.saveToken(1);
}, 1000);

// 2nd request received 2 seconds after start
setTimeout(() => {
  model.request = 2
  model.saveToken(2);
}, 2000);

console.log('Starting - All requests queued');

Prints:

Starting - All requests queued
saveToken for request 1 start - model.request = 1
saveToken for request 2 start - model.request = 2
saveToken for request 1 end - model.request = 2
saveToken for request 2 end - model.request = 2

This was also pointed out in #564 (review).

This will be removed from v3.1 and discussion can continue in #254