Added caching of fort timeouts similar to caching recent forts.

[Quantra]

Short Description:

Caching of fort timeouts similar to how recent_forts are cached and with option to disable this feature.

@douglascamata please review =]

Fixes (provide links to github issues if you can):

-Prevents bot from trying to spin fort still on cool down after restarting bot

[julienlavergne]

Look like it could replace entirely the recent fort cache. right ?

[Quantra]

@anakin5 they perform slightly different tasks. fort_timeouts are all forts you have on cooldown. recent_forts is a list of forts you just been to that you don't want to revisit with user configurable length.

So caching fort_timeouts stops that little bug when you restart next to a fort you just spun.

Caching recent_forts means if a user has a long recent_forts list and restarts their bot regularly then that list isn't started over every time. It ensures the max_circle_size feature works as intended across restarts.

[julienlavergne]

hum, OK but you get the timeout of fort when getting the cell info from the API. So why do you need to keep them in a cache ?
If the spin_fort is spinning a fort on cool down, it is because it is not checking the timeout.

[Quantra]

@anakin5 if you find the answer please let me know! I've been scratching my head on this one....

recent_forts if used by MoveToFort not SpinFort so let's ignore it for now.

When SpinFort works the first thing it does is call self.get_forts_in_range()

    def get_forts_in_range(self):
        forts = self.bot.get_forts(order_by_distance=True)

        for fort in forts:
            if 'cooldown_complete_timestamp_ms' in fort:
                self.bot.fort_timeouts[fort["id"]] = fort['cooldown_complete_timestamp_ms']
                forts.remove(fort)

        forts = filter(lambda fort: fort["id"] not in self.bot.fort_timeouts, forts)
        forts = filter(lambda fort: distance(
            self.bot.position[0],
            self.bot.position[1],
            fort['latitude'],
            fort['longitude']
        ) <= Constants.MAX_DISTANCE_FORT_IS_REACHABLE, forts)

        return forts

That first for loop looks like it throws out anything with a timeout to me but it doesn't work right after a restart.

So my ideas...

Maybe that for loop is badly coded and forts should be cloned before looping... Pretty sure it's NOT ok to be removing things from what you are iterating over whilst you are iterating over it. I could be wrong though.

      for fort in forts.copy():
            if 'cooldown_complete_timestamp_ms' in fort:
                self.bot.fort_timeouts[fort["id"]] = fort['cooldown_complete_timestamp_ms']
                forts.remove(fort)

OR We're not storing the timeouts data correctly.

OR The server isn't returning the timeouts right after relogging.

[julienlavergne]

The server is returning the timeout, but at the start of the bot, the fort_timeouts list is not filled.
The fort_timeouts list is only filled after spinning a fort, that is why you have an issue when restarting the bot.

The first part of the solution, I think, is to fill the fort_timeouts list when the bot start. Which should be easy thing since we get the list of forts in the get_meta_cell function.

After done with that, this loop in spin_forts can be removed because there will never be any fort on timeout that is not in the fort_timeouts list.

Last thing, once the 2 previous thing are done, is to fix the behavior when the inventory is full. When spinning a fort and getting the inventory full response, that fort is not added to the fort_timeout list. This cause the bot to loop over forts when inventory is full.

[Quantra]

I've not looked too closely at what requests the bot makes at start up but if you can find where it gets that data and pop it in fort_timeouts it sounds like it would make things cleaner.

Cacheing the timeouts on exit shouldn't be necessary but I'll leave this PR open for the discussion.

I'll test the forloop fix if i get some time later even if it's just a stop gap fix at least it's a 1 line stop gap fix ;)

[TheSavior]

@Quantra I'm going to close this to reduce clutter, but feel free to open a new PR when you guys are happy with the approach.

We calculate the meta cell here, perhaps this is what you are looking for: https://github.com/PokemonGoF/PokemonGo-Bot/blob/master/pokemongo_bot/__init__.py#L485