GetMapObjects cells contain no forts or pokemon

[aquintero]

This just started happening today. I think something changed on niantic's end.

Edit: looks to be an issue with protobuf structure

AeonLucid/POGOProtos#131

[progmatik]

Hi, did they change their API ?

[RGunning]

looks like it. Few other repos are also reporting issues mchristopher/PokemonGo-DesktopMap#621

[itsfolf]

As I said in https://github.com/PokemonGoF/PokemonGo-Bot-Desktop/issues/65

Well, pokemon go was released in Brazil, and the servers are unstable now, maybe thats the error.
Or with the brazil release, there was also an update that broke bots and apps again, (trackemon.com was woorking too and not working more).

[obamaphone]

What does it mean, change API? Why cant this app simply be updated to the new API (whatever that is)?

[ChipWolf]

@obamaphone most of these apps are using a third-party API which has been put together by reverse engineering PoGO itself. When Niantic decides to change things, it's a fairly complicated process to understand what and how exactly those things were changed. Re-implementation is a different issue entirely.

[cicatiello]

i'll copy myself hoping it helps clarifying problem source PokemonGoF/PokemonGo-Bot#2540:

i tried a little debug on the demo.py and it seems more keyed to direct unmarshalling from pogoprotos;

Line 173,demo.py - 2016-08-03 23:13:33,900 - INFO - Sorting Nearest Forts:
s2_cell_id: 1385709364483981312
current_timestamp_ms: 1470258814176

[]
s2_cell_id: 1385709366631464960
current_timestamp_ms: 1470258814176

[]

the [] are the unmarshalled forts, i.e. empty, i.e. cells do not have forts anymore

[p0psicles]

Niantic made a change in their api. There detecting if the request came from a mobile device. All trackers / bots are affected.

[joeljacobson]

Indeed, they have changed it, I noticed errors last night, and wrote another bot using another project - this is also affected. Boo :-(

[polybahn]

well, not likely changed the api. If they changed the network protocol, then the client part should be updated too, however, this did not happen.

[Dracon23]

They seem to have enabled server side verification of something the client was sending for a while already.

-------- Ursprüngliche Nachricht --------
Von: Yeyao Zhang notifications@github.com
Gesendet: Thursday, August 4, 2016 11:13 AM
An: rubenvereecken/pokemongo-api pokemongo-api@noreply.github.com
Betreff: Re: [rubenvereecken/pokemongo-api] GetMapObjects cells contain no forts or pokemon (#68)

well, not likely changed the api. If they changed the network protocol, then the client part should be updated too, however, this did not happen.

You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHubhttps://github.com//issues/68#issuecomment-237497848, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AFl29_X6QflmerkepifwFumivjrVc_Q_ks5qca0hgaJpZM4JcJSD.

[Mixone-FinallyHere]

Niantics Labs (the company of Pokemon Go) has made changes to how the internal API works (the system of communications between server and phone apps).
This bot is based on a API that just last night was modified by the company, hence why it doesn't work.
One of the main changes we have discovered (the Internet has discovered) is the Maps Response framework. This is why you get 0 pokestops. Any and all errors at this point are not unexpected, we do not know the exact extent of the recent changes so in general no Bot has been able to keep up with this update.

[Mixone-FinallyHere]

sorry I just copy pasted from a comment I made on a different Bot issue so some of the things might be irrelevant to discussion, main idea holds though
the change is apparently somewhere in the Maps Response framework where it is possible names have changed and other smaller things (hashes are assumed to still be correct since we can still do login and get inventory info through bots with old API)

[KazukiOnodera]

http://www.ownedcore.com/forums/pokemon-go/pokemon-go-hacks-cheats/568584-pokemon-go-api-change-reason-not-working-bots-maps.html

[progmatik]

did they resolved the issue ?
http://i.imgur.com/9Zy1ooK.png

edit : https://github.com/pkmngodev/Unknown6/wiki
a question of time before the protobuff will be resolved

[dnsBlah]

please close.

[dnsBlah]

!!! There is a new type of ban, which seems... temporary.. (still)
It may happen you won't find ANY objects in the cells. Out of the blue.

After a while, it starts working again.

I confirm this by using multiple bots, all have the same issue, when I go to VPN or use proxy, it starts working again. Switch back, same issue.

After a while, it works again.

[itsfolf]

@dnsBlah That's a softban.

[dnsBlah]

@checkplys
It has nothing to do with the normal softban. With a softban you will see mapobjects, and you can encounter even spin.
But you won't retrieve items and every pokemon will flee away.
I'm checking this, and run to the nearest pokestop to get rid of this. ;-)

This type of ban is in since last night.

Even with a complete inactive account, you have the same symptons.

[IamCarbonMan]

This issue has already been solved by almost every other major API, see keyphact/pgoapi. It has to do with a new encryption on RequestEnvelope, took those working on it over at POGOProtos about 5 days to fix. Requires a native encryption library to be installed and a small change to be made to request envelope building.

[dmadisetti]

@IamCarbonMan already implemented

[IamCarbonMan]

Still getting a "no forts found" error on running demo.py, exactly like it was a week ago before unknown6 was discovered.

[dmadisetti]

With master? Demo makes a lot more calls than the keyphact one, upping delay times works pretty well. Checkout trainer https://github.com/rubenvereecken/pokemongo-api/tree/trainer

if you're still having issues. About to merge this in, should have best chance

[IamCarbonMan]

The trainer branch currently doesn't work at all, no demo to run.

[dmadisetti]

It's been pulled out into root

[IamCarbonMan]

Oh, obviously. Sorry about that, however demo.py in the trainer branch does have exactly the same behavior.

[dmadisetti]

Try solution to #87 ?

[IamCarbonMan]

Okay, specifying -e './libencrypt.so' with the ./ fixed it for me. Might be a good idea to make that required in argparser.

[dnsBlah]

thats just weird
i think the specification is os related as well

[dnsBlah]

@dmasisetti I found a other project that supplies all variants
shouldnt we provide then all and let python choose the correct version?
Or apply in setup script
with notification that your not responsible etc blablabla

[IamCarbonMan]

Providing compiled encrypt libraries is dangerous since they're made with copyrighted code. However, autodetecting the library by os should be trivial to implement.

[dnsBlah]

guess so ye
lets go on the deep web 😆