Add download queue data button

[jackieo5023]

This PR:

• Adds a button and endpoint to download course queue data as a csv on the course page
  • This is only viewable to course staff, which I have tested
• Changed a comment that was probably not changed after copy/paste to say "remove" instead of "add"

Here is an example of what the download would look like given fake data. Unfortunately, I couldn't upload to GitHub as a csv.
Here is the SQL query that this was based off of:

SELECT
  questions.id AS id,
  courses.name AS CourseName,
  queues.name AS QueueName,
  u_asked.netid AS AskedBy_netid, u_asked.universityName AS AskedBy_RealName,
  u_answered.netid AS AnsweredBy_netid, u_answered.universityName AS AnsweredBy_RealName,
  topic,
  CONVERT_TZ(enqueueTime, '+0:00', 'US/Central') AS enqueueTime,
  CONVERT_TZ(dequeueTime, '+0:00', 'US/Central') AS dequeueTime,
  CONVERT_TZ(answerStartTime, '+0:00', 'US/Central') AS answerStartTime,
  CONVERT_TZ(answerFinishTime, '+0:00', 'US/Central') AS answerFinishTime,
  comments, preparedness,
  questions.location AS UserLocation,
  queues.location AS QueueLocation,
  CONVERT_TZ(queues.createdAt, '+0:00', 'US/Central') AS Queue_CreatedAt,
  queueId, courseId
FROM questions
  INNER JOIN queues ON questions.queueId = queues.id
  INNER JOIN courses ON queues.courseId = courses.id
  INNER JOIN users u_asked ON questions.askedById = u_asked.id
  LEFT JOIN users u_answered ON questions.answeredById = u_answered.id
WHERE courseId=6
ORDER BY enqueueTime DESC

[vercel]

This pull request is being automatically deployed with ZEIT Now (learn more).
To see the status of your deployment, click below or on the icon next to each commit.

🔍 Inspect: https://zeit.co/illinois/queue/fmpfyvbdci
🌍 Preview: In Progress

[nwalters512]

Sweet! It'll be great to have these data exports be self-service instead of requiring manual queries from an admin. I left a few inline comments. A few broader questions:

• Is there a reason the download is associated with the course as a whole and not with a specific queue? Now that queues are being used "persistently", I would imagine it would be most useful to have this data be per-queue. That's not to say an endpoint to get data for a course as a whole wouldn't be useful too!
• To my point of potentially wanting more downloads in the future, it might be nice to design for that from the start and put downloads in a location other than the course header. Perhaps a separate "downloads" section at the bottom of the course page?

[james9909]

Adding onto Nathan's comments, we should also add some tests for this new endpoint

[nwalters512]

Calling something a "real" name has certain unwanted connotations. For instance, I have trans friends whose university name is not their real name; there are also students who use a different name day-to-day from what's in Enterprise. I think there are a couple potential changes here:

• calling this UniversityName
• the above, but also exporting the PreferredName field
• doing what the UI/API already do: taking the preferred name if it's specified and falling back to the name provided by Shib if they aren't
• omitting the name entirely

I think my personal preference would be the third option: it's most consistent with how names are treated everywhere else in the queue. There is the potential for abuse (I can set my preferred name to an arbitrary string), but the nedid can always be cross-referenced with a roster or the university directory if there are doubts about a user's identity.

[jackieo5023]

I'll play around with the options you listed here - maybe one of them can solve the weird issue mentioned above.

[jackieo5023]

@nwalters512 as for your point about downloads being in another section, I think at this point @wadefagen just wants downloads to be pushed out as quick as possible - so I guess think of this as a temporary solution, and later I can make it nicer to be able to download per queue as well.

[jackieo5023]

Now ready for review again.

• Sequelize is weird in that if you specifically include an attribute in a related model, it doesn't exclude other attributes. So in the getColumns function, I whitelist the fields we want to display.
• Added tests for the endpoint.
• RealName is changed to UniversityName
• Moved all logic to api except for the download
• Changed timezone calculation to be on the javascript side rather than sequelize, as sqlite and mysql don't have the same timezone conversion functions

Here is an updated version of an example csv.

[jackieo5023]

Tested on staging. Everything appears to be working properly and correctly (times are verified to be in CST as well).

[nwalters512]

Going to give this a final look over!

[nwalters512]

Thought: this may be undesirable if/when we roll out to schools like UBC that aren't in the central timezone. When that happens, we should either a) report times in UTC and allow consumers of this data to transform it as needed, or b) associate some concept of timezone with either courses or institutions (one institutions are an entity that we have). I think I'd prefer the latter; at any rate, that's outside the scope of this PR.

[nwalters512]

This is looking really good! Merging is blocked on fixing the href on the download button; the other few comments are just questions.

[nwalters512]

Woot, looks great! 🚀