MockRedis

MockRedis provides the same interface as redis-rb, but it stores its data in memory instead of talking to a Redis server. It is intended for use in tests.

Ruby 3.x redis-rb 5.x

The current implementation is tested against Redis 6.2 and 7.0. Older versions may work, but can also return different results or not support some commands.

Install the gem:

gem install mock_redis

It's as easy as require 'mock_redis'; mr = MockRedis.new. Then you can call the same methods on it as you can call on a real Redis object.

For example:

require 'mock_redis'
mr = MockRedis.new
mr.set('some key', 'some value') # => "OK"
mr.get('some key') # => "some value"

mock_redis supports most of the methods that redis-rb does. Examples of supported methods:

• String methods: get, set, append, incr, etc.
• List methods: lpush, lpop, lrange, rpoplpush, etc.
• Set methods: sadd, sinter, sismember, srandmember, etc.
• Hash methods: hset, hget, hgetall, hmget, hincrby, hincrbyfloat etc.
• Sorted set methods: zadd, zrank, zunionstore, etc.
• Expirations: expire, pexpire, ttl, pttl, etc.
• Transactions: multi, exec, discard
• Futures

A MockRedis object can't do everything that a real Redis client can since it's an in-memory object confined to a single process. MockRedis makes every attempt to be Redis-compatible, but there are some necessary exceptions.

• Blocking list commands (#blmove, #blpop, #brpop, and #brpoplpush) work as expected if there is data for them to retrieve. If you use one of these commands with a nonzero timeout and there is no data for it to retrieve, then the command returns immediately. However, if you ask one of these commands for data with a 0 timeout (means "wait forever") and there is no data available, then a MockRedis::WouldBlock exception is raised. It's not what a real Redis client would do, but it beats hanging your test run forever.

• #info just returns canned values; they don't update over time.

• #sort supports ascending and descending sort. ALPHA sort is not yet supported.

Some stuff, we just can't do with a single Ruby object in a single Ruby process.

• Debugging commands (#debug('object', key) and #debug('segfault')) aren't available.

• #object isn't available since we don't have any Redis internals to poke at.

• #monitor isn't available; there's no place for requests to come from, so there's nothing to receive.

• Pubsub commands (#psubscribe, #publish, #punsubscribe) aren't available.

• #slowlog isn't available.

• Scripting commands (#script, #eval, #evalsha) are just stubs—they won't execute anything

There are some things we want to have in here, but that we just haven't gotten to doing yet. If you're interested in helping out, please submit a pull request with your (tested!) implementation.

• #config(:get|:set|:resetstat) isn't done. They can just return canned values.

We recommend running Redis within a Docker container to make development as simple as possible, but as long as you have a Redis instance listening on localhost:6379 you should be good to go.

1. Start Redis.

   docker run --rm -p 6379:6379 redis:6.2-alpine

2. Install dependencies.

   bundle install

3. Run tests.

   bundle exec rspec

These tests were written with Redis running on localhost without any passwords required. If you're using a different version of Redis, you may see failures due to error message text being different and other breaking changes over time.

If you're interested in seeing the changes and bug fixes between each version of mock_redis, read the MockRedis Changelog.

This project is released under the MIT license.