we're still reliant on a lot of hardcoding from the registrations repository. let's get that up to speed.

we're starting to pass session objects around like crazy. can we not do this and use scoped_session from sqlalchemy instead? https://docs.sqlalchemy.org/en/latest/orm/contextual.html

week 8 results sample:
{
"name": "Actors: Songs Connection",
"total_points": 4787,
"watching": 2256,
"completed": 0,
"dropped": 685,
"score": 5.98,
"favorites": 24,
"forum_posts": 11
}

actual results:
Actors: Songs Connection
Episode discussion posts (21): 525
Users watching: 2273
In 19 or fewer active teams: watching points x2
Users dropped (704): -4224
Favorites (24): 1200
Total points: 2047

why is total points so far off?

does it make sense that we keep going back to the config file for things like number of actives, number of bench, min weeks between swaps?

shouldn't these be fixed when the season is created? if the config file changes for season info we'd have some pretty undefined behavior going on.

Double anime's "currently watching" and "completed" points if 4% or fewer of the teams have included it in their active anime.

Currently, it's at 3%.

now that jikan has added that functionality, we should make use of it (once jikanpy supports thenew feature)

maybe we need to use a different date parsing library?

python run_fal.py --collect-series
Traceback (most recent call last):
  File "run_fal.py", line 19, in <module>
    from fal.controllers.load_teams import team_ages, load_teams
  File "/mnt/c/Users/kei-clone/workspace/fantasy-anime-league/fal/controllers/load_teams.py", line 6, in <module>
    from dateutil.parser import parse
  File "/home/kei-clone/.local/share/virtualenvs/fantasy-anime-league-hKlN2DrN/lib/python3.8/site-packages/dateutil/parser/__init__.py", line 2, in <module>
    from ._parser import parse, parser, parserinfo, ParserError
  File "/home/kei-clone/.local/share/virtualenvs/fantasy-anime-league-hKlN2DrN/lib/python3.8/site-packages/dateutil/parser/_parser.py", line 42, in <module>
    import six
ModuleNotFoundError: No module named 'six'

TODO here: https://github.com/fantasy-anime-league/fantasy-anime-league/blob/master/fal/clients/mfalncfm_main.py#L20

Currently, check_usernames.py relies on a regex search of the MAL profile pages in order to determine whether a username is an actual MAL profile. This is a very brittle process that has already been broken and will break every time MAL changes anything significant in their DOM for the profile pages.

Currently, collect_series.py relies on manually generating the advanced search url by hand in order to find the series for the season. This is a very brittle process that has already been broken and will break every time MAL changes up their search path.

double the currently watcing + completed score of an anime if it's owned by 3% of teams or less (configurable percentage)

To simplify the workflow and be more aligned with the general structure of a python package, we should have a single entry point to all functionality in the FAL engine. Most likely, fal.py can be the entry point with a main that uses argparse from the command line to determine which script to run for which feature. This would be preferred over running different disparate scripts, especially from the src directory.

Change from 25, 50, 75, 75

forum posts weekly based on anime and team performance

currently we use a combination of requests and regex to get the bench swaps from MAL's forum posts. we should use something less brittle like Beautiful Soup instead.

possibly create a generative adversarial network alongside it

Let's make this object oriented as originally intended and get us ready to leverage our mysql resources.

or does sqlalchemy autopopulate the ids?