如何在Map/Reduce索引中统计特定字段值的文档数量

现有文档包含activated字段（值为true或false），示例文档如下：

{
"userId": "user1",
"activated": true,
"loginDuration":  123
}

已创建的基础Map/Reduce索引函数如下：
Map函数：

from user in docs.Users
select new {
    user.userId,
    Count = 1,
    user.loginDuration,
    AvgLoginDuration = 0
}

Reduce函数：

from r in results
group r by new r.userId into g
let totalLoginDuration = g.Sum(x => x.loginDuration)
let count = g.Sum(x => x.Count)
select new
{
    userId = g.Key.userId,
    Count = count,
    TotalLoginDuration = totalLoginDuration,
    AvgLoginDuration = totalLoginDuration / count
}

尝试添加统计activated为true的文档数功能时遇到两种问题：

1. 第一种方法无法保存：

// map函数
from user in docs.Users
select new {
    user.userId,
    user.activated,
    Count = 1,
    user.loginDuration,
    AvgLoginDuration = 0,
    NumActivated = 0
}

// reduce函数
from r in results
group r by new r.userId into g
let totalLoginDuration = g.Sum(x => x.loginDuration)
let count = g.Sum(x => x.Count)
select new
{
    userId = g.Key.userId,
    activated = g.Key.activated,
    Count = count,
    TotalLoginDuration = totalLoginDuration,
    AvgLoginDuration = totalLoginDuration / count,
    NumActivated = g.Sum(x => x.activated == true)
}

2. 第二种方法可保存但运行时报错（无法将bool转为decimal）：

// map函数
from user in docs.Users
select new {
    user.userId,
    Count = 1,
    user.loginDuration,
    AvgLoginDuration = 0,
    NumActivated = user.activated == true ? 1 : 0
}

// reduce函数
from r in results
group r by new r.userId into g
let totalLoginDuration = g.Sum(x => x.loginDuration)
let count = g.Sum(x => x.Count)
select new
{
    userId = g.Key.userId,
    Count = count,
    TotalLoginDuration = totalLoginDuration,
    AvgLoginDuration = totalLoginDuration / count,
    NumActivated = g.Sum(x => x.NumActivated == 1)
}

错误原因分析 #

• 第一种方法：分组键仅包含userId，却在结果中错误引用g.Key.activated；同时Sum(x => x.activated == true)返回布尔值，而Sum方法要求输入数值类型，导致无法保存。
• 第二种方法：Reduce函数中Sum(x => x.NumActivated == 1)的判断返回布尔值，Sum需要累加数值，因此触发“无法将bool转为decimal”的错误。

正确的Map/Reduce实现 #

Map函数：
在Map阶段将activated的布尔值转换为数值（true为1，false为0），确保后续Reduce可直接累加：

from user in docs.Users
select new {
    user.userId,
    Count = 1,
    user.loginDuration,
    AvgLoginDuration = 0,
    NumActivated = user.activated ? 1 : 0
}

Reduce函数：
直接对Map阶段生成的NumActivated字段求和，无需额外判断：

from r in results
group r by new r.userId into g
let totalLoginDuration = g.Sum(x => x.loginDuration)
let count = g.Sum(x => x.Count)
let activatedCount = g.Sum(x => x.NumActivated)
select new
{
    userId = g.Key.userId,
    Count = count,
    TotalLoginDuration = totalLoginDuration,
    AvgLoginDuration = totalLoginDuration / count,
    NumActivated = activatedCount
}

说明 #

• Map阶段将布尔值转为数值，保证Reduce阶段的Sum操作能正常累加统计数量。
• Reduce阶段直接对NumActivated求和，即可得到当前userId下activated为true的文档总数。

内容的提问来源于stack exchange，提问作者dbblackdiamond