CODEWITHSUNDEEP
rediskeyredis2.8
Itay Moav -Malimovka
Itay Moav -Malimovka

Reputation: 53603

How to restrict a scan to a specific group of keys

In redis (2.8.x) I have three entity types (I define the entities).
For simplicity, let's call them organizations, departments, users.
When I run scan, it iterates on all keys in the memory, but brings only what matches the match clause.
Is there a way to restrict the scan to only a subset of key? For example search only the users group of keys?
Each group has a different (but consistent) key name pattern (org:[id] dep:[id] user:[id])

Upvotes: 1

Views: 774

Answers (1)

Ofir Luzon
Ofir Luzon

Reputation: 10917

SCAN has an optional MATCH argument that accepts a glob pattern.

You can do something like this to scan through your users objects:

SCAN 0 MATCH user:*

Edit as it is too long for a comment

Clearly scan is not a magic method, a complete cycle (until you receive "0" cursor) must go through all of your keyspace. The MATCH option does the filtering in the Redis side instead of doing it in your client.

As Itamar suggested in his comment, to be able to get the keys you need without going thorough the complete keyspace, you must index them in sets (You must also maintain it on removals).

For example, if your users keys are:

user:31 user31data
user:45 user45data
user:67 user67data
user:13 user13data

You should have a SET of user IDs:

users:IDs {31, 45, 67, 13}

To retrieve all users data, and you probably wants both the ID and the data:

sort users:IDs by nosort get # get user:*

Upvotes: 2

Related Questions