ps:以前都在iteye写博文,现在工作换了,转战前端,基本告别了java和python,看这边氛围还不错,就把那里的博客迁移过来了~~~
出于对性能的要求,公司希望把Mysql的数据迁移到MongoDB上,于是我开始学习Mongo的一些CRUD操作,由于第一次接触NoSQL,还是有点不习惯。
先吐个槽,公司的Mongo版本是2.6.4,而用的java驱动包版本是超级老物2.4版。当时一个“如何对分组后的文档进行筛选”这个需求头痛了很久,虽然shell命令下可以使用Aggregation很方便地解决,但是java驱动包从2.9.0版本才开始支持该特性,我至今没有找到不用Aggregation解决上述需求的办法。只能推荐公司升级驱动包版本,希望没有后续的兼容问题。
Mongo2.2版本后开始支持Aggregation Pipeline,而java驱动包从2.9.0版本才开始支持2.2的特性,2.9版本是12年发布的,mongodb在09年就出现了,似乎Mongo对java的开发者不怎么友好←_←
MongoDB目前提供了三个可以执行聚合操作的命令:aggregate、mapReduce、group。三者在性能和操作的优劣比较见官网提供的表格 Aggregation Commands Comparison,这里不再赘述细节。
aggregate、mapReduce、group原型及内部实现
我从官网总结出来了这三个函数的原型及底层封装的命令
函数名:db.collection.group()
函数原型:
db.collection.group(
{
key,
reduce,
initial
[, keyf]
[, cond]
[, finalize]
}
)
封装的命令:
db.runCommand(
{
group:
{
ns: <namespace>,
key: <key>,
$reduce: <reduce function>,
$keyf: <key function>,
cond: <query>,
finalize: <finalize function>
}
}
)
函数名:db.collection.mapReduce()
函数原型:
db.collection.mapReduce(
<map>,
<reduce>,
{
out: <collection>,
query: <document>,
sort: <document>,
limit: <number>,
finalize: <function>,
scope: <document>,
jsMode: <boolean>,
verbose: <boolean>
}
)
封装的命令:
db.runCommand(
{
mapReduce: <collection>,
map: <function>,
reduce: <function>,
finalize: <function>,
out: <output>,
query: <document>,
sort: <document>,
limit: <number>,
scope: <document>,
jsMode: <boolean>,
verbose: <boolean>
}
)
函数名:db.collection.aggregate()
函数原型:
db.collection.aggregate(
pipeline,
options
)
封装的命令:
db.runCommand(
{
aggregate: "<collection>",
pipeline: [ <stage>, <...> ],
explain: <boolean>,
allowDiskUse: <boolean>,
cursor: <document>
}
)
Mysql与MongoDB对聚合处理的对比
好记性不如烂笔头,下面通过操作来了解这几个函数和命令
1、准备测试数据
先准备SQL的测试数据(用来验证结果、比较SQL语句和NoSQL的异同):
先创建数据库表:
create table dogroup (
_id int,
name varchar(45),
course varchar(45),
score int,
gender int,
primary key(_id)
);
插入数据:
insert into dogroup (_id, name, course, score, gender) values (1, "N", "C", 5, 0);
insert into dogroup (_id, name, course, score, gender) values (2, "N", "O", 4, 0);
insert into dogroup (_id, name, course, score, gender) values (3, "A", "C", 5, 1);
insert into dogroup (_id, name, course, score, gender) values (4, "A", "O", 6, 1);
insert into dogroup (_id, name, course, score, gender) values (5, "A", "U", 8, 1);
insert into dogroup (_id, name, course, score, gender) values (6, "A", "R", 8, 1);
insert into dogroup (_id, name, course, score, gender) values (7, "A", "S", 7, 1);
insert into dogroup (_id, name, course, score, gender) values (8, "M", "C", 4, 0);
insert into dogroup (_id, name, course, score, gender) values (9, "M", "U", 7, 0);
insert into dogroup (_id, name, course, score, gender) values (10, "E", "C", 7, 1);
接着准备MongoDB测试数据:
创建Collection(等同于SQL中的表,该行可以不写,Mongo会在插入数据时自动创建Collection)
db.createCollection("dogroup")
插入数据:
db.dogroup.insert({"_id": 1,"name": "N",course: "C","score": 5,gender: 0})
db.dogroup.insert({"_id": 2,"name": "N",course: "O","score": 4,gender: 0})
db.dogroup.insert({"_id": 3,"name": "A",course: "C","score": 5,gender: 1})
db.dogroup.insert({"_id": 4,"name": "A",course: "O","score": 6,gender: 1})
db.dogroup.insert({"_id": 5,"name": "A",course: "U","score": 8,gender: 1})
db.dogroup.insert({"_id": 6,"name": "A",course: "R","score": 8,gender: 1})
db.dogroup.insert({"_id": 7,"name": "A",course: "S","score": 7,gender: 1})
db.dogroup.insert({"_id": 8,"name": "M",course: "C","score": 4,gender: 0})
db.dogroup.insert({"_id": 9,"name": "M",course: "U","score": 7,gender: 0})
db.dogroup.insert({"_id": 10,"name": "E",course: "C","score": 7,gender: 1})
以下操作可能逻辑上没有实际意义,主要是帮助熟悉指令
2、查询每门课程参与考试的人数
SQL写法:
select course as '课程名', count(*) as '数量' from dogroup group by course;
MongoDB写法:
① group方式
db.dogroup.group({
key : { course: 1 },
initial : { count: 0 },
reduce : function Reduce(curr, result) {
result.count += 1;
},
finalize : function Finalize(out) {
return {"课程名": out.course, "数量": out.count};
}
});
返回的格式如下:
{
"课程名" : "C",
"数量" : 4
},
{
"课程名" : "O",
"数量" : 2
},
{
"课程名" : "U",
"数量" : 2
},
{
"课程名" : "R",
"数量" : 1
},
{
"课程名" : "S",
"数量" : 1
}
② mapReduce方式
db.dogroup.mapReduce(
function () {
emit(
this.course,
{course: this.course, count: 1}
);
},
function (key, values) {
var count = 0;
values.forEach(function(val) {
count += val.count;
});
return {course: key, count: count};
},
{
out: { inline : 1 },
finalize: function (key, reduced) {
return {"课程名": reduced.course, "数量": reduced.count};
}
}
)
这里把count初始化为1的原因是,MongoDB执行完map函数(第一个函数)后,如果key所对应的values数组的元素个数只有一个,reduce函数(第二个函数)将不会被调用。
返回的格式如下:
{
"_id" : "C",
"value" : {
"课程名" : "C",
"数量" : 4
}
},
{
"_id" : "O",
"value" : {
"课程名" : "O",
"数量" : 2
}
},
{
"_id" : "R",
"value" : {
"课程名" : "R",
"数量" : 1
}
},
{
"_id" : "S",
"value" : {
"课程名" : "S",
"数量" : 1
}
},
{
"_id" : "U",
"value" : {
"课程名" : "U",
"数量" : 2
}
}
③ aggregate方式
db.dogroup.aggregate(
{
$group:
{
_id: "$course",
"数量": { $sum: 1 }
}
}
)
返回格式如下:
{ "_id" : "S", "数量" : 1 }
{ "_id" : "R", "数量" : 1 }
{ "_id" : "U", "数量" : 2 }
{ "_id" : "O", "数量" : 2 }
{ "_id" : "C", "数量" : 4 }
以上三种方式中,group得到了我们想要的结果,mapReduce返回的结果只能嵌套在values里面,aggregate必须返回_id,无法为分组的字段指定别名,但是无疑第三种是最简单的。
虽然上面的问题不影响程序在前台展现数据,但是对于一个略微有强迫症的开发者确实难以忍受的。本人才疏学浅,刚接触Mongo,不知道后两者有没有可行的方法来获取想要的结果,希望网友指教。
3、查询Docouments(等同于SQL中记录)数大于2的课程
SQL写法:
select course, count(*) as count from dogroup group by course having count > 2;
MongoDB写法:
① aggregate方式(注意$group和$match的先后顺序)
db.dogroup.aggregate({
$group: {
_id: "$course",
count: { $sum: 1 }
}
},{
$match: {
count:{
$gt: 2
}
}
});
目前尚未找到group和mapReduce对分组结果进行筛选的方法,欢迎网友补充
4、找出所有分数高于5分的考生数量及分数,返回的格式为“分数、数量”
SQL写法:
select score as '分数', count(distinct(name)) as '数量' from dogroup where score > 5 group by score;
MongoDB写法:
① group方式
db.dogroup.group({
key : { score: 1 },
cond : { score: {$gt: 5} },
initial : { name:[] },
reduce : function Reduce(curr, result) {
var flag = true;
for(i=0;i<result.name.length&&flag;i++){
if(curr.name==result.name[i]){
flag = false;
}
}
// 如果result.name数组里面没有curr.name则添加curr.name
if(flag){
result.name.push(curr.name);
}
},
finalize : function Finalize(out) {
return {"分数": out.score, "数量": out.name.length};
}
});
② mapReduce方式
db.dogroup.mapReduce(
function () {
if(this.score > 5){
emit(
this.score,
{score: this.score, name: this.name}
);
}
},
function (key, values) {
var reduced = {score: key, names: []};
var json = {};//利用json对象的key去重
for(i = 0; i < values.length; i++){
if(!json[values[i].name]){
reduced.names.push(values[i].name);
json[values[i].name] = 1;
}
}
return reduced;
},
{
out: { inline : 1 },
finalize: function (key, reduced) {
return {"分数": reduced.score, "数量": reduced.names?reduced.names.length:1};
}
}
)
③ aggregate方式
db.dogroup.aggregate({
$match: {
score: {
$gt: 5
}
}
},{
$group: {
_id: {
score: "$score",
name: "$name"
}
}
},{
$group: {
_id: {
"分数": "$_id.score"
},
"数量": { $sum: 1 }
}
});
弄熟上面这几个方法,大部分的分组应用场景应该没大问题了。
这张图示可以更直观地理解(点击看大图):