Reimplement /archive

[bdc34]

Reproduces /archive route for browse.

To try it out:
clone the repo, check out the branch,
pipenv install
pipenv shell
FLASK_APP=app.py FLASK_DEBUG=1 pipenv run flask run
Then go to some archives
Single category archive:
http://localhost:5000/archive/hep-ph
Multi category archive:
http://localhost:5000/archive/physics
Multi category archive with descriptions:
http://localhost:5000/archive/astro-ph
Subsumed category:
http://localhost:5000/archive/acc-phys

[bdc34]

Note: there is a failing test that is also failing on develop. I'm currently ignoring that as it goes by in the test results but we could wait until it is fixed in develop.
This test is fixed now.

[mhl10]

Looks great, with minor changes requested. You had mentioned a couple of "stray" sub-pages that we agreed shouldn't be part of this work. Do you remember what those were? Just want to record them somewhere.

[bdc34]

Looks great, with minor changes requested. You had mentioned a couple of "stray" sub-pages that we agreed shouldn't be part of this work. Do you remember what those were? Just want to record them somewhere.

Looks like:
Catch up controller.
Year controller.
Find controller.

[bbarker]

Code looks good to me, just a few minor questions/suggestions.

The archive pages look relatively simple enough that we probably don't need to repurpose our original /abs comparison script to be run on this, but I thought I'd throw it out there as an idea. The other aspect of it would be that maybe it doesn't matter as much if everything stays (almost) the same in /archive as it does in /abs.

[bbarker]

Just to confirm I know what is happening here, I think Python is flattening the elements of what is inside the [ ] to be a list. But it looks like this almost a reverse-map style of syntax, where the output is specified first ((id, ARCHIVES[id]["name"])) then the list of things to map over is specified (the filtered ids). Would be good to know what this is. Also, are archives just List[str, str]? might be nice to annotate to make it a little more clear

[bdc34]

You have it. The reverse-map syntax is python list comprehension.
I often use flatten (as in flatmap) to mean [[a],[b,c]] => [a,b,c] which isn't happening here. But I think you mean flaten in a metaphorical sense.

[bdc34]

@bbarker

The archive pages look relatively simple enough that we probably don't need to repurpose our original /abs comparison script to be run on this, but I thought I'd throw it out there as an idea. The other aspect of it would be that maybe it doesn't matter as much if everything stays (almost) the same in /archive as it does in /abs.

I was thinking of doing something like that. I thought it would be useful to gather several thousands requests from the access logs and test them against NG /archive. It would also be good to test NG /list this way.

I was thinking that this could happen during the testing phase of this release.