Aerospike 分组统计

2015-2-5 18:37:9
收藏：0
阅读：70
评论：1

我将 Aerospike 设置为以下结构： bin: item，timestamp，qty，price，discount

我正在尝试实现以下等效 SQL：

SELECT item, sum(qty), sum(price), sum(discount)
FROM items
WHERE timestamp between 20150101000000 and 20150101235959 GROUP BY item

我尝试编写以下 Lua 脚本，但在执行脚本时会出现以下错误：

错误：（1）AEROSPIKE_ERR_SERVER：“UDF：执行错误 2：/opt/aerospike /usr/udf/lua/aggregateByItem.lua:21：尝试索引空值”

有人能帮助解决这个问题吗？

local function aggregate_item_stats(itemMap,rec)
  local item = rec.item
  local innerMap = itemMap[item]
  if innerMap == nil then
        innerMap = {qty = 0, price = 0, discount = 0}
  end
  innerMap.qty = innerMap.qty + (tonumber(rec.qty) or 0);
  innerMap.price = innerMap.price + (tonumber(rec.price) or 0);
  innerMap.discount = innerMap.discount + (tonumber(rec.discount) or 0);
  itemMap[item] = innerMap
  return itemMap
end

local function fn_merge(a,b)
  a.qty = a.qty + b.qty
  a.price = a.price + b.price
  a.discount = a.discount + b.discount
  return a
end
local function reduce_values(a, b)
  return map.merge(a, b, fn_merge)
end
function highUsageReport(stream)
 return stream : aggregate(itemMap{}, aggregate_item_stats) : reduce(reduce_values)
end
点赞
用户1261267
用户1261267

工作代码如下，主要改动在第4行：

if innerMap == null then

修改为：

if innerMap == nil then

local function aggregate_item_stats(itemMap, rec)
 local key = rec.item
 local innerMap = itemMap[key]
 if innerMap == nil then
  innerMap = map {qty = 0, price = 0, discount = 0}
 end
 innerMap.qty = (tonumber(rec.qty) or 0)
 innerMap.price = innerMap.price + tonumber(rec.price or 0)
 innerMap.discount = innerMap.discount + tonumber(rec.discount or 0)
 itemMap[key] = innerMap
 return itemMap
end

local function reduce_values(a, b)
 return map.merge(a, b, fn_merge)
end

local function fn_merge(a, b)
 a.qty = tonumber(a.qty) + tonumber(b.qty)
 a.price = tonumber(a.price) + tonumber(b.price)
 a.discount = tonumber(a.discount) + tonumber(b.discount)
 return a
end

function highUsageReport(stream)
 return stream : aggregate(map(), aggregate_item_stats)
    : reduce(reduce_values)
end
2015-02-05 19:25:46

作者:

用户1261267

用户1261267
推荐文章