IMPORTANT: always use SCAN
instead of (the evil) KEYS
![do not use <code>KEYS</code>](https://i.stack.imgur.com/9D6XA.jpg)
Redis' pattern matching is somewhat functionally limited (see the implementation of stringmatchlen
in util.c) and does not provide that which you seek ATM. That said, consider the following possible routes:
- Extend
stringmatchlen
to match your requirements, possibly submitting it as a PR.
- Consider what you're trying to do - fetching a subset of keys is always going to be inefficient unless you index them, consider tracking the names of all non-user keys (i.e.g. in a Redis Set) instead.
- If you are really insistent on scanning the entire keyspace and match against negative patterns, one way to accomplish that is with a little bit of Lua magic.
Consider the following dataset and script:
127.0.0.1:6379> dbsize
(integer) 0
127.0.0.1:6379> set user:1 1
OK
127.0.0.1:6379> set use:the:force luke
OK
127.0.0.1:6379> set non:user a
OK
Lua (save this as scanregex.lua
):
local re = ARGV[1]
local nt = ARGV[2]
local cur = 0
local rep = {}
local tmp
if not re then
re = ".*"
end
repeat
tmp = redis.call("SCAN", cur, "MATCH", "*")
cur = tonumber(tmp[1])
if tmp[2] then
for k, v in pairs(tmp[2]) do
local fi = v:find(re)
if (fi and not nt) or (not fi and nt) then
rep[#rep+1] = v
end
end
end
until cur == 0
return rep
Output - first time regular matching, 2nd time the complement:
foo@bar:~$ redis-cli --eval scanregex.lua , "^user"
1) "user:1"
foo@bar:~$ redis-cli --eval scanregex.lua , "^user" 1
1) "use:the:force"
2) "non:user"
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…