Question

先看一下官方的调用示例代码：

local redis = require "resty.redis"
local red = redis:new()

red:set_timeout(1000) -- 1 sec

local ok, err = red:connect("127.0.0.1", 6379)
if not ok then
    ngx.say("failed to connect: ", err)
    return
end

ok, err = red:set("dog", "an animal")
if not ok then
    ngx.say("failed to set dog: ", err)
    return
end

ngx.say("set result: ", ok)

-- put it into the connection pool of size 100,
-- with 10 seconds max idle time
local ok, err = red:set_keepalive(10000, 100)
if not ok then
    ngx.say("failed to set keepalive: ", err)
    return
end

这是一个标准的 Redis 接口调用，如果你的代码中 Redis 被调用频率不高，那么这段代码不会有任何问题。但如果你的项目重度依赖 Redis，工程中有大量的代码在重复创建连接-->数据操作-->关闭连接（或放到连接池）这个完整的链路调用完毕，甚至还要考虑不同的 return 情况做不同处理，就很快发现代码中有大量的重复。

Lua 是不支持面向对象的。很多人用尽各种招术利用元表来模拟。可是，Lua 的发明者似乎不想看到这样的情形，因为他们把取长度的 __len 方法以及析构函数 __gc 留给了 C API，纯 Lua 只能望洋兴叹。

我们期望的代码应该是这样的：

local red = redis:new()
local ok, err = red:set("dog", "an animal")
if not ok then
    ngx.say("failed to set dog: ", err)
    return
end

ngx.say("set result: ", ok)

local res, err = red:get("dog")
if not res then
    ngx.say("failed to get dog: ", err)
    return
end

if res == ngx.null then
    ngx.say("dog not found.")
    return
end

ngx.say("dog: ", res)

期望它自身具备以下几个特征：

• new、connect 函数合体，使用时只负责申请，尽量少关心什么时候具体连接、释放；
• 默认 Redis 数据库连接地址，但是允许自定义；
• 每次 Redis 使用完毕，自动释放 Redis 连接到连接池供其他请求复用；
• 要支持 Redis 的重要优化手段 pipeline；

不卖关子，只要干货，我们最后是这样干的，可以这里看到 gist代码

-- file name: resty/redis_iresty.lua
local redis_c = require "resty.redis"


local ok, new_tab = pcall(require, "table.new")
if not ok or type(new_tab) ~= "function" then
    new_tab = function (narr, nrec) return {} end
end


local _M = new_tab(0, 155)
_M._VERSION = '0.01'


local commands = {
    "append",            "auth",              "bgrewriteaof",
    "bgsave",            "bitcount",          "bitop",
    "blpop",             "brpop",
    "brpoplpush",        "client",            "config",
    "dbsize",
    "debug",             "decr",              "decrby",
    "del",               "discard",           "dump",
    "echo",
    "eval",              "exec",              "exists",
    "expire",            "expireat",          "flushall",
    "flushdb",           "get",               "getbit",
    "getrange",          "getset",            "hdel",
    "hexists",           "hget",              "hgetall",
    "hincrby",           "hincrbyfloat",      "hkeys",
    "hlen",
    "hmget",              "hmset",      "hscan",
    "hset",
    "hsetnx",            "hvals",             "incr",
    "incrby",            "incrbyfloat",       "info",
    "keys",
    "lastsave",          "lindex",            "linsert",
    "llen",              "lpop",              "lpush",
    "lpushx",            "lrange",            "lrem",
    "lset",              "ltrim",             "mget",
    "migrate",
    "monitor",           "move",              "mset",
    "msetnx",            "multi",             "object",
    "persist",           "pexpire",           "pexpireat",
    "ping",              "psetex",            "psubscribe",
    "pttl",
    "publish",      --[[ "punsubscribe", ]]   "pubsub",
    "quit",
    "randomkey",         "rename",            "renamenx",
    "restore",
    "rpop",              "rpoplpush",         "rpush",
    "rpushx",            "sadd",              "save",
    "scan",              "scard",             "script",
    "sdiff",             "sdiffstore",
    "select",            "set",               "setbit",
    "setex",             "setnx",             "setrange",
    "shutdown",          "sinter",            "sinterstore",
    "sismember",         "slaveof",           "slowlog",
    "smembers",          "smove",             "sort",
    "spop",              "srandmember",       "srem",
    "sscan",
    "strlen",       --[[ "subscribe",  ]]     "sunion",
    "sunionstore",       "sync",              "time",
    "ttl",
    "type",         --[[ "unsubscribe", ]]    "unwatch",
    "watch",             "zadd",              "zcard",
    "zcount",            "zincrby",           "zinterstore",
    "zrange",            "zrangebyscore",     "zrank",
    "zrem",              "zremrangebyrank",   "zremrangebyscore",
    "zrevrange",         "zrevrangebyscore",  "zrevrank",
    "zscan",
    "zscore",            "zunionstore",       "evalsha"
}


local mt = { __index = _M }


local function is_redis_null( res )
    if type(res) == "table" then
        for k,v in pairs(res) do
            if v ~= ngx.null then
                return false
            end
        end
        return true
    elseif res == ngx.null then
        return true
    elseif res == nil then
        return true
    end

    return false
end


-- change connect address as you need
function _M.connect_mod( self, redis )
    redis:set_timeout(self.timeout)
    return redis:connect("127.0.0.1", 6379)
end


function _M.set_keepalive_mod( redis )
    -- put it into the connection pool of size 100, with 60 seconds max idle time
    return redis:set_keepalive(60000, 1000)
end


function _M.init_pipeline( self )
    self._reqs = {}
end


function _M.commit_pipeline( self )
    local reqs = self._reqs

    if nil == reqs or 0 == #reqs then
        return {}, "no pipeline"
    else
        self._reqs = nil
    end

    local redis, err = redis_c:new()
    if not redis then
        return nil, err
    end

    local ok, err = self:connect_mod(redis)
    if not ok then
        return {}, err
    end

    redis:init_pipeline()
    for _, vals in ipairs(reqs) do
        local fun = redis[vals[1]]
        table.remove(vals , 1)

        fun(redis, unpack(vals))
    end

    local results, err = redis:commit_pipeline()
    if not results or err then
        return {}, err
    end

    if is_redis_null(results) then
        results = {}
        ngx.log(ngx.WARN, "is null")
    end
    -- table.remove (results , 1)

    self.set_keepalive_mod(redis)

    for i,value in ipairs(results) do
        if is_redis_null(value) then
            results[i] = nil
        end
    end

    return results, err
end


function _M.subscribe( self, channel )
    local redis, err = redis_c:new()
    if not redis then
        return nil, err
    end

    local ok, err = self:connect_mod(redis)
    if not ok or err then
        return nil, err
    end

    local res, err = redis:subscribe(channel)
    if not res then
        return nil, err
    end

    res, err = redis:read_reply()
    if not res then
        return nil, err
    end

    redis:unsubscribe(channel)
    self.set_keepalive_mod(redis)

    return res, err
end


local function do_command(self, cmd, ... )
    if self._reqs then
        table.insert(self._reqs, {cmd, ...})
        return
    end

    local redis, err = redis_c:new()
    if not redis then
        return nil, err
    end

    local ok, err = self:connect_mod(redis)
    if not ok or err then
        return nil, err
    end

    local fun = redis[cmd]
    local result, err = fun(redis, ...)
    if not result or err then
        -- ngx.log(ngx.ERR, "pipeline result:", result, " err:", err)
        return nil, err
    end

    if is_redis_null(result) then
        result = nil
    end

    self.set_keepalive_mod(redis)

    return result, err
end


for i = 1, #commands do
    local cmd = commands[i]
    _M[cmd] =
            function (self, ...)
                return do_command(self, cmd, ...)
            end
end


function _M.new(self, opts)
    opts = opts or {}
    local timeout = (opts.timeout and opts.timeout * 1000) or 1000
    local db_index= opts.db_index or 0

    return setmetatable({
            timeout = timeout,
            db_index = db_index,
            _reqs = nil }, mt)
end


return _M

调用示例代码：

local redis = require "resty.redis_iresty"
local red = redis:new()

local ok, err = red:set("dog", "an animal")
if not ok then
    ngx.say("failed to set dog: ", err)
    return
end

ngx.say("set result: ", ok)

在最终的示例代码中看到，所有的连接创建、销毁连接、连接池部分，都被完美隐藏了，我们只需要业务就可以了。妈妈再也不用担心我把 Redis 搞垮了。