mongodb高级操作(1)-update

前面我们已经介绍了mongodb的基本操作:mongodb基本操作

下面是介绍一些高级的增删改查操作.

1.更新文档

文档替换

最简单的方法就是文档完全替换,如下就是一个完全替换的例子,先把需要更新的文档找出来并且赋值,然后修改所赋的值,最后再进行update:

>a1=db.test.findOne()

{"_id":ObjectId("5738785d132e1e47e535a177"),"x":3.14,"y":5.5}

>a1

{"_id":ObjectId("5738785d132e1e47e535a177"),"x":3.14,"y":5.5}

>deletea1.x

true

>a1

{"_id":ObjectId("5738785d132e1e47e535a177"),"y":5.5}

>a1.content={name:'job',age:11}

{"name":"job","age":11}

>a1

{

"_id":ObjectId("5738785d132e1e47e535a177"),

"y":5.5,

"content":{

"name":"job",

"age":11

}

}

>db.test.update({"_id":ObjectId("5738785d132e1e47e535a177")},a1)

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

使用修改器

通常文档只会有一部分需要更新,可以使用原子性的更新修改器对指定文档中的某些键进行更新

"$set"修改器

"$set"修改器用来指定一个字段的值,如果这个字段不存在则创建它,例如先给test表添加一个phone属性:

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com"}

>

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$set":{phone:666}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com","phone":666}

然后用修改器再将phone该为888

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$set":{phone:888}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com","phone":888}

使用"$set"修改器还可以修改字段的属性,例如可以把phone改成一个数组,还可以用"$unset"将键完全删除:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$unset":{phone:888}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com"}

使用"$set"修改内嵌文档:

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com"}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":888,"address":"abc"}}

>

>db.test.update({"_id":ObjectId("57395fa85f74882a9bfa2d9f")},{"$set":{"id.phone":666}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com"}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

"$inc"增加和减少

"$inc"修改器用来增加和减少已有键的值,只能用于整形,长整形或者双精度浮点型的值,用在其它类型的值上面会报错.对于不存在的键,像"$set"那样也会自动创建相应的键,并且值为给定的值

例如我们用"$inc"修改器给age值+1,原来是30,现在变成了31

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com"}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$inc":{age:1}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":31,"email":"xxxx@qq.com"}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

如果要减少,则将age:1里的1改成-1

"$push"数组修改器-添加元素

如果数组已经存在,那么"$push"会向已有的数组末尾添加一个元素,如果没有就创建一个新的数组,下面这个例子会创建一个表示comments的数组:

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com","phone":888}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$push":{comments:

...{name:"jack",content:"good"}

...}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com","phone":888,"comments":[{"name":"jack","content":"good"}]}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

继续给comments添加值:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$push":{comments:{name:"tom",content:"verygood"}}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573957c55f74882a9bfa2d9e"),"name":"brent","age":30,"email":"xxxx@qq.com","phone":888,"comments":[{"name":"jack","content":"good"},{"name":"tom","content":"verygood"}]}

{"_id":ObjectId("57395fa85f74882a9bfa2d9f"),"id":{"phone":666,"address":"abc"}}

还可以将"$push"和"$each"一起使用,一次添加多个值

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$push":{comments:{"$each":["a","b","c"]}}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>

>

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"a",

"b",

"c"

]

}

还可以将"$slice"和"$push"组合在一起使用,这样就保证数组不会超出设定好的最大长度,这实际上得到了一个最多包含N个元素的数组:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$push":{comments:{"$each":["a","b","c"],"$slice":-10}}})

这个例子会限制数组只包含最后加入的10个元素,"$slice"的值必须是负整数.

还可以使用"$sort"对数组中的对象进行排序,注意"$slice"必须和"$each"配合使用.

"$ne"和"$addToSet"将数组作为数据集使用

当将数组作为数据集使用进行的更新的时候,要保证数组内的元素不能重复,当update的数据重复即不做,如果不重复则进行更新,可以使用"$ne",也可以使用"$addToSet"

使用"$ne"的方法:

>db.test.update({"comments":{"$ne":"jack"}},{"$push":{"comments":"jack"}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"a",

"b",

"c",

"a",

"b",

"c",

"jack"

]

感觉ne就是notequal的意思

其实按照我的理解上面那句话就是当comment这个数组内的元素ne(不等于)jack的时候,使用push在comment里将jack这个元素添加到commnts数组中.

使用"$addToSet"的方法:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$addToSet":{"comments":"andy"}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"a",

"b",

"c",

"a",

"b",

"c",

"jack",

"andy"

]

}

个人觉得"$addToset"的方法更好用一点,而且格式和上面介绍的集中类似.而且"$addToset"可以和"$each"组合起来进行批量的添加

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$addToSet":{"comments":{"$each":["andy","lily","lucy"]}}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"a",

"b",

"c",

"a",

"b",

"c",

"jack",

"andy",

"lily",

"lucy"

]

}

"$pop"和"$pull"删除数组元素

可以使用"$pop"方法将数组进行出列操作{"$pop:{"key":1}}从数组key的末尾元素开始出列,key为-1从头部开始出列,下面这个例子将lucy删除掉了:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$pop":{"comments":1}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"a",

"b",

"c",

"a",

"b",

"c",

"jack",

"andy",

"lily"

]

}

如果不是从开头或者末尾删除,可以使用"$pull",它会删除所有匹配的文档,如下删除了所有的a:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$pull":{"comments":"a"}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"good"

	},

{

			"name":"tom",

			"content":"verygood"

	},

"b",

"c",

"b",

"c",

"jack",

"andy",

"lily"

]

}

$基于位置的数组修改

如果数组有多个值,而我们只想对其中一部分进行操作,有两种方法操作数组中的值:通过位置或者定位操作符$

数组下标都是以0开始的,可以将下标直接作为键的选择元素,下面这个例子给comments这个数组的第一个元素添加一个num:1的元素

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"tom",

			"content":"verygood"

	},

"jack",

"andy",

"lily"

]

}

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$set":{"comments.0.num":1}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>

>

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"tom",

			"content":"verygood",

			"num":1

	},

"jack",

"andy",

"lily"

]

}

再对这个num加上1:

>db.test.update({"_id":ObjectId("573957c55f74882a9bfa2d9e")},{"$inc":{"comments.0.num":1}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"tom",

			"content":"verygood",

			"num":2

	},

"jack",

"andy",

"lily"

]

}

这种方法的弊端是需要先查询出数据,然后根据数据你才知道需要修改的数组的下标.为了客服这个困难mongodb提供了定位操作符$,用来定位匹配的元素,定位符只匹配第一个的元素,例如我们把上面的tom改成jack:

>db.test.update({"comments.name":"tom"},{"$set":{"comments.$.name":"jack"}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.findOne()

{

	"_id":ObjectId("573957c55f74882a9bfa2d9e"),

	"name":"brent",

	"age":30,

	"email":"xxxx@qq.com",

	"phone":888,

	"comments":[

{

			"name":"jack",

			"content":"verygood",

			"num":2

	},

"jack",

"andy",

"lily"

]

}

upsert

upsert是一种特殊的更新,要是没有找到符合条件的更新那么就创建一个新的文档,如果找到就正常更新.

update的第三个参数即控制update是普通的update还是upsert,默认为false,即普通的update,如果设置成true,则为upsert

下面的这个例子查找name为bob的,如果有就给age加1,如果没有则创建name为bob,并且设置age为1,使用upsert实现:

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":28}

>db.test.update({"name":"bob"},{"$inc":{"age":1}},true)

WriteResult({

	"nMatched":0,

	"nUpserted":1,

	"nModified":0,

	"_id":ObjectId("573c86d3017c5eb7d08aed6d")

})

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":28}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1}

>

有的时候需要创建文档的同时创建字段并为他赋值,但是在之后的更新中这个字段的值不再改变,这就是"$setOnInsert"的作用.

"$setOnInsert"只会在文档插入的时候设置字段的值.

下面这个例子第一次使用upsert,因为没有name为tom的所以插入age为10,第二次再使用upsert想设置tom的age为20,因为使用

"$setOnInsert"而且tom已经存在那么就不会再更新了

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":28}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1}

>db.test.update({"name":"tom"},{"$setOnInsert":{"age":10}},true)

WriteResult({

	"nMatched":0,

	"nUpserted":1,

	"nModified":0,

	"_id":ObjectId("573c88fe017c5eb7d08aed6e")

})

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":28}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1}

{"_id":ObjectId("573c88fe017c5eb7d08aed6e"),"name":"tom","age":10}

>db.test.update({"name":"tom"},{"$setOnInsert":{"age":20}},true)

WriteResult({"nMatched":1,"nUpserted":0,"nModified":0})

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":28}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1}

{"_id":ObjectId("573c88fe017c5eb7d08aed6e"),"name":"tom","age":10}

saveshell帮助程序

save是一个shell函数,如果文档不存在,它会自动创建文档,如果存在它就更新这个文档,他只有一个参数:文档.

如果这个文档中含有_id则会调用upsert,如果没有则调用insert

>vara=db.test.findOne()

>a

{

	"_id":ObjectId("573c858c323f7f2e2ccb0e17"),

	"name":"brent",

	"age":28

}

>a.age=40

40

>a

{

	"_id":ObjectId("573c858c323f7f2e2ccb0e17"),

	"name":"brent",

	"age":40

}

>db.test.save(a)

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":40}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1}

{"_id":ObjectId("573c88fe017c5eb7d08aed6e"),"name":"tom","age":10}

更新多个文档

mongodb默认情况下如果匹配到了多个文档,在更新的时候默认只会更新第一个文档,其他文档不发生变化,如果要更新所有的文档则需要将update的第四个参数设置为true,默认为false

例如下面的例子,第一次更新不设置第四个参数,返回只update了一条数据,第二次设置第四个参数为true,则会匹配更新所有数据:

>db.test.find({"name":"brent"})

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":40}

{"_id":ObjectId("573c8bd3323f7f2e2ccb0e18"),"name":"brent","age":28}

>db.test.update({"name":"brent"},{"$inc":{"age":1}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.update({"name":"brent"},{"$inc":{"age":1}},false,true)

WriteResult({"nMatched":2,"nUpserted":0,"nModified":2})

>db.test.find({"name":"brent"})

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":42}

{"_id":ObjectId("573c8bd3323f7f2e2ccb0e18"),"name":"brent","age":29}

>

如果想知道多文档更新到底更新了多少文档,可以运行getLastError命令,(可以理解为返回最后一次操作的相关信息),建n的值就是被更新的文档的数量:

>db.test.update({"name":"brent"},{"$inc":{"age":1}},false,true)

WriteResult({"nMatched":2,"nUpserted":0,"nModified":2})

>db.runCommand({getLastError:1})

{

	"connectionId":1,

	"updatedExisting":true,

	"n":2,

	"syncMillis":0,

	"writtenTo":null,

	"err":null,

	"ok":1

}

返回被更新的文档

getLastError只能获得更新的相关信息,并不能返回被更新的文档,可以通过findAndModify命令得到被更新的文档.

下面这个例子,查询status为ready的文档,进行处理,处理完成之后将ready改成done,先来看看当前的数据:

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":43,"status":"done"}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1,"status":"ready"}

{"_id":ObjectId("573c88fe017c5eb7d08aed6e"),"name":"tom","age":10,"status":"ready"}

{"_id":ObjectId("573c8bd3323f7f2e2ccb0e18"),"name":"brent","age":30,"status":"done"}

下面这条语句是将集合test中status为ready的一个文档修改为running:

>ps=db.runCommand({"findAndModify":"test",#集合名

..."query":{"status":"ready"},

..."sort":{"age":-1},

..."update":{"$set":{"status":"running"}}}).value

然后你可以对这条数据进行一些处理

Process(ps).......

最后处理完成之后,将这个文档status设置为done

>db.test.update({"_id":ps._id},{"$set":{"status":"done"}})

WriteResult({"nMatched":1,"nUpserted":0,"nModified":1})

>db.test.find()

{"_id":ObjectId("573c858c323f7f2e2ccb0e17"),"name":"brent","age":43,"status":"done"}

{"_id":ObjectId("573c86d3017c5eb7d08aed6d"),"name":"bob","age":1,"status":"ready"}

{"_id":ObjectId("573c88fe017c5eb7d08aed6e"),"name":"tom","age":10,"status":"done"}

{"_id":ObjectId("573c8bd3323f7f2e2ccb0e18"),"name":"brent","age":30,"status":"done"}

这样就提供了一个很好的操作队列和原子性的方法,而且你可以写一个循环把集合中的文档类似于放到一个队列中处理.

findAndModify有很多参数可以使用,例如上面我们使用到了query,sort,update

findAndModify

字符串集合名

query

查询文档,用于检索文档的条件

sort

排序结果集,我们处理的时候可能需要按照某个顺序来进行处理

update

修改器文档,用于对匹配的文件进行修改

remove

布尔类型,表示是否删除文档(remove和update必须指定中的一个)

remove的语法如下:

.....

"remove":true}).value

new

布尔类型,表示返回更新前的文档还是更新后的,默认返回更新前的文档

fields

文档中需要返回的字段(可选)

upsert

布尔类型,当是true的时候表示这是一个upsert,默认为false

update和remove必须有一个,而且只能有一个,如果没有匹配的文档,这个命令会返回一个错误