I have many filters in my leaderboards i.e. around 15. I cannot make separate leaderboard against each filter combination. Is there any way I can filter members according to my choice? The only filter which I was able to find is members in a list. How can I do if I want to find if member is of specific type for example?

It is possible to customize the redis_connection, but what about the connection_pool?

I would like to use the decode_responses from ConnectionPool of redis-py option:

redis.from_url(redis_url, decode_responses=True)

Leaderboard('...', ?????)

Leaderboard('...', redis_connection=redis.from_url(redis_url, decode_responses=True))????
Leaderboard('...', connection=redis.from_url(redis_url, decode_responses=True))????

Should not be a good idea let the decode_responses=True as default in the leaderboard lib?

Thanks in advance.

I'm using leaderboard-python for creating leaderboard.
Global rank is coming out fine but I wish to rank users based on their country.
Do I have to create different leaderboard for each country or is there any filtering mechanism for this?

P.S. I have very little idea of Redis, so explanation with sample code would be very much appreciated.

Use CompetitionLeaderboard and i have a set of friends - some of whom may or may not be part of the Leaderboard. When i call ranked_in_list with the complete list of my friends, it returns incorrect ranks for some of the friends who are part of the Leaderboard.

Hi, I'm seeing this error log when calling rank_member on an existing member.

I should mention this is when running unit tests on a dev machine.

The values instance.user.id and instance.score evaluate to 2 and 0 respectively (both int).

Traceback (most recent call last):
...
  File "<removed>", line 43, in update_leaderboard
    lb.rank_member(instance.user.id, instance.score)
  File "C:\<removed>\site-packages\leaderboard\__init__.py", line 92, in rank_member
    self.rank_member_in(self.leaderboard_name, member, score, member_data)
  File "C:\<removed>\site-packages\leaderboard\__init__.py", line 106, in rank_member_in
    pipeline.execute()
  File "C:\<removed>\site-packages\redis\client.py", line 2154, in execute
    return execute(conn, stack, raise_on_error)
  File "C:\<removed>\site-packages\redis\client.py", line 2069, in _execute_transaction
    self.raise_first_error(commands, response)
  File "C:\<removed>\site-packages\redis\client.py", line 2105, in raise_first_error
    raise r
ResponseError: Command # 1 (ZADD unittesting_leaderboard (-: (DEFAULT: ), 32000) 1) of pipeline caused error: value is not a double

I tried the same command from within the shell (on the same database), and no error occurs. What might be the cause?

Thanks in advance,

I have to show realtime leaderboard for 1,00,000 users , so How can I implement it in my Django webapp ,

If I am insert my data through Loop it will take too much time and load , so Can you please tell me what will the best approach for this
Here you can Check my code
Models.py

`class Score(models.Model):
name=models.CharField(max_length=211)
score=models.IntegerField()

def str(self):
return self.name
`

Views.py

def Index(request):
a=Score.objects.all()
highscore_lb=CompetitionRankingLeaderboard('sdfs')
for index in a:
highscore_lb.rank_member(index.name, index.score)
b=highscore_lb.all_leaders()
return render(request,'index.html',{'b':b})

Hi,

we are using Redis with a single node, its working fine for us.

when we configured the same with cluster primary endpoint, xxxx.xxx.clustercfg.aps1.cache.amazonaws.com we are getting an error like

    628         if isinstance(response, ResponseError):
--> 629             raise response
    630         return response
    631 

ResponseError: MOVED 1896 10.0.0.27:6379

when I am using 10.0.0.27 everything working fine.

but the point here is to use the primary endpoint DNS value, irrespective of the ip, can you suggest us the correct way to use the cluster endpoint

If you pass Leaderboard(connection=redis.StrictRedis()) it will not consider it since you do a type introspection for only redis.Redis in the initialize function.

It took a while for me to notice that you were not re-using my connection so you might want to either raise an error, document the behavior or re-use the connection_pool only.

Hi,
We have decided to use your awesome package to replace our postgres one. During the testing period, we have encountered an issue regarding time-based leaderboards. When I use expire_leaderboard_at on our leaderboard which uses global member data, it sets the same expiration date on the global member data too, which is not what we desire since other keys use this data too.
As I tracked the code, this method "_member_data_key" returns the member data key for "expire_leaderboard_at" which causes the global member data to be deleted.
Is there a workaround to prevent this behavior or is it in your plan to develop this feature?

Doesn't look like this exists in python3.

For ranked_in_list API in CompetitionLeaderboard if any user id doesn't exist in a leaderboard and 'sort_by' = 'rank' is used, then the API throws an exception ( 'rank' key not found ). Whereas it works if the 'sort_by' = 'score' is used.

The bug is 'rank' is not initialised to None - unlike 'score' which by default is initialised to None.
This can be fixed with the following:

data[self.RANK_KEY] = None in line 100 or so as shown here:

        score = responses[index * 2 + 1]
        if score is not None:
            score = float(score)
        data[self.SCORE_KEY] = score

        data[self.RANK_KEY] = None # This is the additional line

        if self.order == self.ASC:
            data[self.RANK_KEY] = self.redis_connection.zcount(
                leaderboard_name, '-inf', "(%s" % str(float(data[self.SCORE_KEY])))
        else:
            #print "SCORE_KEY - " + str(data[self.SCORE_KEY])
            if data[self.SCORE_KEY] != None:
                    data[self.RANK_KEY] = self.redis_connection.zcount(
                        leaderboard_name, "(%s" % str(float(data[self.SCORE_KEY])), '+inf')

When using python 3.4 and requesting a leaderboard with a page_size=1 it returns the incorrect member due to the way python 3 handles division on line 1007 in leaderboard/leaderboard.py. The work around was to cast to int, so that fixed it for any page size.

starting_offset = reverse_rank_for_member - (page_size / 2)

should be

starting_offset = reverse_rank_for_member - int(page_size / 2)

If you install the latest version of Redis Python Client (3.1.0 at this time) you will receive this error when trying to run the sample code and use the rank_member function for the first time:
AttributeError: 'str' object has no attribute 'items'

This is because of breaking changes that occurred between Redis Client v2.x to v3.x
https://github.com/andymccurdy/redis-py#upgrading-from-redis-py-2x-to-30

This specific issue can be solved as follows, but I'm not sure if other parts may have broken:
https://stackoverflow.com/questions/53553009/not-able-to-insert-data-using-zaddsorted-set-in-redis-using-python

As a super quick fix, you can use redis python client v2.10.6

To install that version of redis run the following:
pip install redis==2.10.6

Please provide the article linked.

Is that expected?

I haven't used member_data in python2, so don't know if this is a python3 oddity or expected

TieRankingLeaderboard should take ties_namespace as an optional argument. However,
you get
TypeError: init() got an unexpected keyword argument 'ties_namespace'
because in the TieRankingLeaderboard, you dont pop off ties_namespace before you send it into Leaderboard, and then into Redis.

I have a leaderboard with 2994 members in it.
lb.member_at(0) should return None, I believe, but it returns {'member': '7694', 'score': 2352.43, 'rank': 25}
Accordingly, lb.member_at(-1) returns {'member': '15488', 'score': 2392.103, 'rank': 24}

Is this behaviour normal? Should I control this on my side?

If a leaderboard has multiple rows for the same rank, then the members_from_rank_range does not return correct value. e.g.

'Person 1' with 120 points is ranked 1
'Person 2' with 120 points is ranked 1
'Person 3' with 112 points is ranked 3

Then
a) invoking members_from_rank_range(1, 1) - returns only 1 item ( expecting 2 )
b) invoking members_from_rank_range(2, 2) - returns 1 item ( expecting none )

Unless I missed it in the docs, is there a way to force ranks to be the same when members have the same score?

>>> for index in range(1, 11):
...    highscore_lb.rank_member('member_%s' % index, index)

 >>> highscore_lb.rank_member('member_%s' % 100, 10.0)
 >>> highscore_lb.leaders(1)
 [{'member': 'member_100', 'score': 10.0, 'rank': 1}, {'member': 'member_10', 'score': 10.0, 'rank': 2}, ...]

As you can see, the first two members have an equal score, but different ranks. Is there a way to have them assigned the same rank?

Thanks