I have an ordered set in Redis (I am actually using a python client https://github.com/andymccurdy/redis-py), for example:
zadd myset 1 key1
zadd myset 1 key2
zadd myset 1 key3
zadd myset 0 key4
Note that 3 keys have the same score.
Using ZRANGE, i would like to get the top 2 entries (i.e lowest scores). "key4" will always be the first result as it has a lower value, but I would like the second return value to be randomly selected between the ties: key1,key2,key3. ZRANGE actually returns the keys in the order they are indexed: "keys1" is always my second result:
zrange myset 0 -1 WITHSCORES
1) "key4"
2) "0"
3) "key1"
4) "1"
5) "key2"
6) "1"
7) "key3"
8) "1"
any idea?
thanks,
J.
As kindly requested by Linus G Thiel, here are more details about my usecase:
I would like to use zsets to perform a simple ranking system. I have a list of items, for each one a score representing the relevance of the item. For the cold start of my system, most of the scores will be identical (i.e 0), and I would like to randomly select among the items having the same score. Otherwise I will always return the exact same lexicographic ordering, which will introduce a bias in the system.
The solution you propose, using one specific set for each duplicated score value will work. I will give it a try.
Thanks,
Related
I'm trying to strategies how to calculate position/rank based on score and date
I have a two pieces of data with the same score shown in the image below.
If i run the command
$ZADD usersScores 11 userId6
It will add it above userId5 which is not something I want
Is it possible to add it below userId1 maybe based on date?
This is my goal essentially where i would like to show the position based on the amount and the oldest would be ontop via date
Thanks!
Sorted Set ordering
Sorted sets are ordered by their scores first, and then lexigraphically by their members if the scores are equivalent. So in your case if you added the following keys:
127.0.0.1:6379> ZADD myset 11 user1
(integer) 1
127.0.0.1:6379> ZADD myset 11 user5
(integer) 1
127.0.0.1:6379> ZADD myset 24 user3
(integer) 1
127.0.0.1:6379> zadd myset 54 user2
(integer) 1
127.0.0.1:6379> zadd myset 11 user6
(integer) 1
and then did a ZRANGE, you'd get them in ascending order with the members ordered in alphabetical order:
127.0.0.1:6379> zrange myset 0 -1 WITHSCORES
1) "user1"
2) "11"
3) "user5"
4) "11"
5) "user6"
6) "11"
7) "user3"
8) "24"
9) "user2"
10) "54"
What it looks like, is your GUI is using, or are just passing in the rev argument to ZRANGE under the hood, which orders everything in reverse (scores descending, reverse alphabetical)
127.0.0.1:6379> ZRANGE myset 0 -1 WITHSCORES REV
1) "user2"
2) "54"
3) "user3"
4) "24"
5) "user6"
6) "11"
7) "user5"
8) "11"
9) "user1"
10) "11"
Second sort-by
There are a couple of ways you could handle the second sort by with a sorted set, let's assume you wanted to do this with a 32 bit unix timestamp
You could prepend the second sortby field to the member name so if user6's score came in at 1659579650, you'd just make the member name be 1659579650:user1, then when it orders it, it would do so lexigraphically
You could combine the score with the second, since 11 is the score, we could shift 11 32 bits to the left and add/or the timestamp to it, which comes out to 48,904,219,906, those will then order correctly, and then when you want to find the score, all you need to do is right-shift the score by 32 bits.
I'm using Redis sorted set to implement the leaderboard of my game, where I show the user ranking in descending order. I'm stuck in a case where two or more users have the same score. So in this case, I want the higher ranking of the user who gets the score first. For example, I'm adding the following entries in Redis.
127.0.0.1:6379> zadd testing-key 5 a
(integer) 1
127.0.0.1:6379> zadd testing-key 4 b
(integer) 1
127.0.0.1:6379> zadd testing-key 5 c
(integer) 1
and when I'm querying for the rank in reverse order, I'm getting this
127.0.0.1:6379> zrevrange testing-key 0 10
1) "c"
2) "a"
3) "b"
but in my case, the ranking should be like
1) "a"
2) "c"
3) "b"
So is there any provision in Redis to give higher precedence to the entity which entered first in the set with the same score?
I found one solution to this problem. In my case, the score is an integer so I converted it into decimal and added Long.MAX_VALUE - System.nanoTime() after decimal. So the final score code will be like
double finalScore = score.(Long.MAX_VALUE - System.nanoTime());
So the final score of the player who scored first would be higher than the second one. Please let me know if you have any better solution.
If your leaderboard's scores are "small" enough, you may get away with using a combination of the score and the timestamp (e.g. 123.111455234, where 123 is the score). However, since the Sorted Set score is a double floating point, you may lose precision.
Alternatively, keep two Sorted Sets - one with each player's leaderboard score and the other with each player's score timestamp, and use both to determine the order.
Or, use a single sorted set for the leader board, encode the timestamp as part of the member and rely on lexicographical ordering.
I'm tracking members in multiple Sorted Sets in Redis as a way to do multi-column indexing on the members. As an example, let's say I have two Sorted Sets, lastseen (which is epoch time) and points, and I store usernames as members in these Sorted Sets.
I'm wanting to use ZREVRANGEBYSCORE against lastseen to get a list of users who were recently seen, but I only want a user to be included in the results if its also present in the points Sorted Set.
I looked at using ZINTERSTORE to help me do this, but the available AGGREGATE options are not conducive to my scenario. If one of the AGGREGATE options was to use the score from a specific set provided to the ZINTERSTORE command, then it would work, but that's not an option.
---EDIT---
Example:
redis> ZADD lastseen 12345 "foo"
redis> ZADD lastseen 12346 "bar"
redis> ZADD lastseen 12347 "sucka"
redis> ZADD points 5 "foo"
redis> ZADD points 9 "bar"
Now, if I run:
redis> ZREVRANGEBYSCORE lastseen +inf -inf
1) "sucka"
2) "bar"
3) "foo"
What I'm looking for is a way to only get "foo" and "bar" returned by the call to ZREVRANGEBYSCORE for lastseen since "sucka" isn't resent in the points Sorted Set.
I can do:
redis> ZINTERSTORE out 2 lastseen points
redis> ZRANGE out 0 -1 WITHSCORES
1) "foo"
2) "12350"
3) "bar"
4) "12355"
As you can see, the ZINTERSTORE does cut out the "sucka" member, but the scores for the other two members get aggregated using SUM. There's two other aggregation options (MIN and MAX) available for use with ZINTERSTORE. In this case, using MAX would give me what I'm looking for, but that wouldn't be the case if a user's points were greater than the epoch value for the last time they were seen.
You can get what you're asking by setting the WEIGHT of points to 0 and of lastseen to 1 when running ZINTERSTORE so that only the weight of lastseen is considered:
redis> ZINTERSTORE out 2 lastseen points WEIGHTS 1 0
(integer) 2
Then, depending on what ordering you're going for, you can either:
redis> ZRANGE out 0 -1 WITHSCORES
1) "foo"
2) "12345"
3) "bar"
4) "12346"
or:
redis> ZREVRANGE out 0 -1 WITHSCORES
1) "bar"
2) "12346"
3) "foo"
4) "12345"
How do I get most weighted elements from a sorted set, but excluding those found in another set(or list or hash).
>zadd all 1 one
>zadd all 2 two
>zadd all 3 three
>sadd disabled 2
>sdiff all disabled
(error) WRONGTYPE Operation against a key holding the wrong kind of value
Is my only option is to get elements from the sorted set one-by-one and compare to the list of "disabled" items? Wouldn't that be very slow because of so many transactions to a server?
What is the approach here?
Note: I assume you've meant sadd disabled two
As you've found out, SDIFF does not operate on sorted sets - that is because defining the difference between sorted sets isn't trivial.
What you could do is first create a temporary set with ZUNIONSTORE and set the intersect's scores to 0. Then do a range excluding the 0, e.g.:
127.0.0.1:6379> ZADD all 1 one 2 two 3 three
(integer) 3
127.0.0.1:6379> SADD disabled two
(integer) 1
127.0.0.1:6379> ZUNIONSTORE tmp 2 all disabled WEIGHTS 1 0 AGGREGATE MIN
(integer) 3
127.0.0.1:6379> ZREVRANGEBYSCORE tmp +inf 1 WITHSCORES
1) "three"
2) "3"
3) "one"
4) "1"
Assume I have a set (or sorted set or list if that would be better) A of 100 to 1000 strings.
Then I have a sorted set B of many more strings, say one million.
Now C should be the intersection of A and B (of the strings of course).
I want to have every tuple (X, SCORE_OF_X_IN_B) where X is in C.
Any Idea?
I got two ideas:
Interstore
store A a sorted set with every score being 0
interstore to D
get every item of D
delete D
Simple loop in client
loop over A in my client programm
get zscore for every string
While 1. has way too much overhead on the redis side (Has to write for example. The redis page states quite a high time complexity, too http://redis.io/commands/zinterstore), 2. would have |A| database connections and won't be a good choice.
Maybe I could write a redis/lua script which will work like zscore but with an arbitrary number of strings, but I'm not sure if my hoster allows scripts...
So I just wanted to ask SO, if there is an elegant and fast solution available without scripting!
There is a simple solution to your problem: ZINTERSTORE will work with a SET and a ZSET. Try:
redis> sadd foo a
(integer) 1
redis> zadd bar 1 a
(integer) 1
redis> zadd bar 2 b
(integer) 1
redis> zinterstore baz 2 foo bar AGGREGATE MAX
(integer) 1
redis> zrange baz 0 -1 withscores
1) "a"
2) "1"
Edit: I added AGGREGATE MAX above, since redis will give each member of the (non-sorted) set foo a default score of 1, and SUM that with whatever score it has in the (sorted) set bar.