TestingParty08: User password reset

First stab at this. This patch relies on the new assertions and API introduced in #296001: Capture e-mails sent during tests and add e-mail assertions / API.

This tests:

1. the password reset functionality by requesting a new password by username
2. the password reset functionality by requesting a new password by e-mail
3. requesting a password for a user that does not exist
4. it verifies that the e-mails go out (and don't go out for #3)
5. extracts the password link from those e-mails.
6. "clicks" the one-time login link
7. verifies that the user is granted access and can change his/her password.
8. Also ensures that the one-time link works only once.

Some of the logic for this test was already present in the existing user registration test, but a work-around was used to create the one-time login link (direct call to user_pass_reset_url instead of capturing the e-mail).

Leaving at NW until #296001: Capture e-mails sent during tests and add e-mail assertions / API is commited, and because the tests sometimes fail for reasons unknown to me.

#296001: Capture e-mails sent during tests and add e-mail assertions / API has been commited, let's take this one for a spin...

The original code to test the password reset URL had the following code:

sleep(1); // TODO Find better way.

When I leave this out, tests succeed on my local computer about 25% of the time. If I put it back in, tests consistently pass, so putting it back in here to see what the testbot says. I browsed the repository but since this was part of the initial commit, I couldn't find any further reasoning as to why it's in there.

Note to reviewers: I haven't figured out why, but the test fails 1 out of 5 runs. The other 4 test runs succeed without any problem...

That sleep(1) is also confusing me.

@catch: thanks for pointing me in the right direction. The current implementation for the one-time login link compares timestamps in seconds, and uses < and >. This means it might take up to one second before the link actually becomes active.

Since I think these links should be active immediately, I've created a separate issue for this: #481794: One-time link should be active immediately. Once this lands, the sleep(1) hack can be removed.

re-rolled.

The sleep(1) is still in patch #12.

Oops... sorry about that.

Re-rolled without the sleep(1)...

Add two missing periods.

Verified that sleep() is not in the patch. Tests installed and reviewed.
User tests pass with no fails - 699 passes, 0 fails, and 0 exceptions

Tests don't qualify as critical.

Patch no longer applies after #46149: Prevent account cancellation for uid 1 was commited.

Fix in D8 first, then backport.

Patch looks good and I was able to apply it. I'll leave this issue in "needs review" for others to test.

I've:
- applied the patch
- run tests few times (php core/scripts/run-tests.sh --file core/modules/user/lib/Drupal/user/Tests/UserPasswordResetTest.php)
- review the code of the test
Everything works fine. I would apply the patch.

I applied the patch, ran the test and got this message:

An AJAX HTTP error occurred. HTTP Result Code: 200 Debugging information follows. Path: /batch?id=5&op=do_nojs&op=do StatusText: OK ResponseText: Exception: The configuration directory type <em class="placeholder">staging</em> does not exist. in config_get_config_directory() (line 526 of /Users/lukehodge/Sites/d8/www/core/includes/bootstrap.inc).

Test works perfectly for me - marking RTBC

Backported #23 to D7.

un assigning since it had been a while and someone else has posted patch anyway. :)

29: 293487-29-password-reset-test.patch queued for re-testing.