MongoDB：3. Schema Design

MongoDB：3. Schema Design

数据库正则表达式schemaschemadesignmongodb

摘要：1. Document-Oriented MongoDB 是一种面向文档(document-oriented)的数据库，其内存储的是一种 JSON-like 结构化数据。尽管拥有和关系型数据库 Database/Table 类似的的 DB/Collection 概念，但同一 Collection 内的
1. Document-Oriented
MongoDB 是一种面向文档(document-oriented)的数据库，其内存储的是一种 JSON-like 结构化数据。尽管拥有和关系型数据库 Database/Table 类似的的 DB/Collection 概念，但同一 Collection 内的 Document 可以拥有不同的属性。
(注: 以下 > 提示符表示 mongo JS 代码，>>> 为 Python 代码)
> use blog 
switched to db blog 

> db.users.insert({name:"user1", age:15}) 
> db.users.insert({name:"user2", age:20, sex:1}) 

> db.users.find() 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "name" : "user1", "age" : 15 } 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "name" : "user2", "age" : 20, "sex" : 1 }
可以通过 $exists 判断某个字段是否存在。
> db.users.find({sex:{$exists:true}}) 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "name" : "user2", "age" : 20, "sex" : 1 }
2. Embed vs. Reference
Document 采取 JSON-like 这种层级结构，因此我们可以直接用嵌入(Embed)代替传统关系型数据库的关联引用(Reference)。
> u = db.users.findOne({name:"user1"}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15 
} 

> u.address = ["address1", "address2"] 
[ "address1", "address2" ] 

> db.users.save(u) 

> db.users.findOne({name:"user1"}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15, 
        "address" : [ 
                "address1", 
                "address2"
        ] 
} 

> db.users.findOne({address:"address1"}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15, 
        "address" : [ 
                "address1", 
                "address2"
        ] 
}
MongoDB 支持以 "." 分割的 namespace 路径，但需要注意 key 不能以 "$" 开头，不能包含 "." 字符 (条件表达式中的多级路径须用引号)。
> u = db.users.findOne({address:"address1"}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15, 
        "address" : [ 
                "address1", 
                "address2"
        ] 
} 

> u.im = {msn:"[e-mail]", qq:12345678} 
{ "msn" : "[e-mail]", "qq" : 12345678 } 

> db.users.save(u) 

> u = db.users.findOne({"im.qq":12345678}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15, 
        "address" : [ 
                "address1", 
                "address2"
        ], 
        "im" : { 
                "msn" : "[e-mail]", 
                "qq" : 12345678 
        } 
} 

> u.im.qq 
12345678 

> u.im.msn 
[e-mail] 

> db.users.update({"im.qq":12345678}, {$set:{"im.qq":12345}}) 

> u = db.users.findOne({"im.qq":12345}) 
{ 
        "_id" : ObjectId("4c479885089df9b53474170a"), 
        "name" : "user1", 
        "age" : 15, 
        "address" : [ 
                "address1", 
                "address2"
        ], 
        "im" : { 
                "msn" : "[e-mail]", 
                "qq" : 12345 
        } 
} 

> u = db.users.find({"im.qq":{$exists:true}}, {"im.qq":1}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "im" : { "qq" : 12345 } }
注意: 由于每篇文档都包含完整的 key 数据，因此使用尽可能短的 key 可以有效节省存储空间。
3. Operator
(1) $all: 判断数组属性是否包含全部条件。
> db.users.insert({name:"user3", data:[1,2,3,4,5,6,7]}) 
> db.users.insert({name:"user4", data:[1,2,3]}) 

> db.users.find({data:{$all:[2,3,4]}}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "name" : "user3", "data" : [ 1, 2, 3, 4, 5, 6, 7 ] }
注意和 $in 的区别。$in 是检查目标属性值是条件表达式中的一员，而 $all 则要求属性值包含全部条件元素。
(2) $size: 匹配数组属性元素数量。
> db.users.insert({name:"user3", data:[1,2,3,4,5,6,7]}) 
> db.users.insert({name:"user4", data:[1,2,3]}) 

> db.users.find({data:{$size:3}}) 
{ "_id" : ObjectId("4c47a13bb48cde79c6780df1"), "name" : "user4", "data" : [ 1, 2, 3 ] }
(3) $type: 判断属性类型。
> db.users.insert({name:"user5", t:1}) 
> db.users.insert({name:"user6", t:1.34}) 
> db.users.insert({name:"user7", t:"abc"}) 

> db.users.find({t:{$type:1}}) 
{ "_id" : ObjectId("4c47a231b48cde79c6780df2"), "name" : "user5", "t" : 1 } 
{ "_id" : ObjectId("4c47a23eb48cde79c6780df3"), "name" : "user6", "t" : 1.34 } 

> db.users.find({t:{$type:2}}) 
{ "_id" : ObjectId("4c47a258b48cde79c6780df4"), "name" : "user7", "t" : "abc" }
类型值:
•double:1
•string: 2
•object: 3
•array: 4
•binary data: 5
•object id: 7
•boolean: 8
•date: 9
•null: 10
•regular expression: 11
•javascript code: 13
•symbol: 14
•javascript code with scope: 15
•32-bit integer: 16
•timestamp: 17
•64-bit integer: 18
•min key: 255
•max key: 127
(4) 使用正则表达式进行查询。
> db.users.find({name:/user[135]/i}, {name:1}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "name" : "user1" } 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "name" : "user3" } 
{ "_id" : ObjectId("4c47a231b48cde79c6780df2"), "name" : "user5" }
>>> users = db.users.find({"name" : {"$regex" : r"(?i)user[135]"}}, ["name"]) 
>>> for u in users: print u 
... 
{u'_id': ObjectId('4c479885089df9b53474170a'), u'name': u'user1'} 
{u'_id': ObjectId('4c47a133b48cde79c6780df0'), u'name': u'user3'} 
{u'_id': ObjectId('4c47a231b48cde79c6780df2'), u'name': u'user5'}
正则表达式标记:
i: 忽略大小写。
m: 默认为单行处理，此标记表示多行。
x: 扩展。
(5) 数组属性元素值匹配。
> db.users.find({data:"abc"}) 
{ "_id" : ObjectId("4c47a481b48cde79c6780df5"), "name" : "user8", "data" : [ { "a" : 1, "b" : 10 }, 3, "abc" ] } 

> db.users.find({data:{$elemMatch:{a:1, b:{$gt:5}}}}) 
{ "_id" : ObjectId("4c47a481b48cde79c6780df5"), "name" : "user8", "data" : [ { "a" : 1, "b" : 10 }, 3, "abc" ] }
{data:"abc"} 仅简单匹配数组属性是否包含该元素。$elemMatch 则可以处理更复杂的元素查找条件。当然也可以写成如下方式。
> db.users.find({"data.a":1, "data.b":{$gt:5}}) 
{ "_id" : ObjectId("4c47a481b48cde79c6780df5"), "name" : "user8", "data" : [ { "a" : 1, "b" : 10 }, 3, "abc" ] }
还可以直接使用序号进行操作。
> db.users.find({"data.2":3}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "name" : "user3", "data" : [ 1, 2, 3, 4, 5, 6, 7 ] } 
{ "_id" : ObjectId("4c47a13bb48cde79c6780df1"), "name" : "user4", "data" : [ 1, 2, 3 ] }
(6) $not: 取反，表示返回条件不成立的文档。
似乎只能跟正则和 $mod 一起使用？？？？
> u = db.users.find({name:{$not:/user3/}}, {name:1}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "name" : "user1" } 
{ "_id" : ObjectId("4c47a13bb48cde79c6780df1"), "name" : "user4" } 
{ "_id" : ObjectId("4c47a231b48cde79c6780df2"), "name" : "user5" } 
{ "_id" : ObjectId("4c47a23eb48cde79c6780df3"), "name" : "user6" } 
{ "_id" : ObjectId("4c47a258b48cde79c6780df4"), "name" : "user7" } 
{ "_id" : ObjectId("4c47a481b48cde79c6780df5"), "name" : "user8" } 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "name" : "user2" }
(7) $unset: 和 $set 相反，表示移除文档属性。
> u = db.users.find({name:"user1"}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "name" : "user1", "age" : 15, "address" : [ "address1", "address2" ], "im" : { "msn" : "[e-mail]", "qq" : 12345 } } 

> db.users.update({name:"user1"}, {$unset:{address:1, im:1}}) 

> u = db.users.find({name:"user1"}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "age" : 15, "name" : "user1" }
(8) $push: 和 $ pushAll 都是向数组属性添加元素。
> u = db.users.find({name:"user1"}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "age" : 15, "name" : "user1" } 

> db.users.update({name:"user1"}, {$push:{data:1}}) 

> u = db.users.find({name:"user1"}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "age" : 15, "data" : [ 1 ], "name" : "user1" } 

> db.users.update({name:"user1"}, {$pushAll:{data:[2,3,4,5]}}) 

> u = db.users.find({name:"user1"}) 
{ "_id" : ObjectId("4c479885089df9b53474170a"), "age" : 15, "data" : [ 1, 2, 3, 4, 5 ], "name" : "user1" }
(9) $addToSet: 和 $push 类似，不过仅在该元素不存在时才添加 (Set 表示不重复元素集合)。
> db.users.update({name:"user2"}, {$addToSet:{data:1}}) 
> db.users.update({name:"user2"}, {$addToSet:{data:1}}) 

> u = db.users.find({name:"user2"}) 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "data" : [ 1 ], "name" : "user2" } 
> db.users.update({name:"user2"}, {$push:{data:1}}) 

> u = db.users.find({name:"user2"}) 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "data" : [ 1, 1 ], "name" : "user2" }
要添加多个元素，使用 $each。
> db.users.update({name:"user2"}, {$addToSet:{data:{$each:[1,2,3,4]}}}) 

> u = db.users.find({name:"user2"}) 
{ "_id" : ObjectId("4c479896089df9b53474170b"), "data" : [ 1, 2, 3, 4 ], "name" : "user2" }
(10) $pop: 移除数组属性的元素，$pull 按值移除，$pullAll 移除所有符合提交的元素。
> u = db.users.find({name:"user3"}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "data" : [ 1, 2, 3, 4, 5, 6, 7, 2, 3 ], "name" : "user3" } 

> db.users.update({name:"user3"}, {$pop:{data:1}}) // 移除最后一个元素 

> u = db.users.find({name:"user3"}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "data" : [ 1, 2, 3, 4, 5, 6, 7, 2 ], "name" : "user3" } 

> db.users.update({name:"user3"}, {$pop:{data:-1}}) // 移除第一个元素 

> u = db.users.find({name:"user3"}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "data" : [ 2, 3, 4, 5, 6, 7, 2 ], "name" : "user3" } 

> db.users.update({name:"user3"}, {$pull:{data:2}}) // 移除全部 2 

> u = db.users.find({name:"user3"}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "data" : [ 3, 4, 5, 6, 7 ], "name" : "user3" } 

> db.users.update({name:"user3"}, {$pullAll:{data:[4,5,6]}}) // 移除 4,5,6 

> u = db.users.find({name:"user3"}) 
{ "_id" : ObjectId("4c47a133b48cde79c6780df0"), "data" : [ 3, 7 ], "name" : "user3" }
(11) $where: 用 JS 代码来代替有些丑陋的 $lt、$gt。
MongoDB 内置了 Javascript Engine (SpiderMonkey)。可直接使用 JS Expression，甚至使用 JS Function 写更复杂的 Code Block。
> for (var i = 0; i < 10; i++) db.users.insert({name:"user"+i, age:i}) 

> db.users.find() 
{ "_id" : ObjectId("4c47b3372a9b2be866da226e"), "name" : "user0", "age" : 0 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da226f"), "name" : "user1", "age" : 1 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2270"), "name" : "user2", "age" : 2 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2271"), "name" : "user3", "age" : 3 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2272"), "name" : "user4", "age" : 4 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2273"), "name" : "user5", "age" : 5 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2274"), "name" : "user6", "age" : 6 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2275"), "name" : "user7", "age" : 7 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2276"), "name" : "user8", "age" : 8 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2277"), "name" : "user9", "age" : 9 } 

> db.users.find({$where:"this.age > 7 || this.age < 3"}) 
{ "_id" : ObjectId("4c47b3372a9b2be866da226e"), "name" : "user0", "age" : 0 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da226f"), "name" : "user1", "age" : 1 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2270"), "name" : "user2", "age" : 2 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2276"), "name" : "user8", "age" : 8 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2277"), "name" : "user9", "age" : 9 } 

> db.users.find("this.age > 7 || this.age < 3") 
{ "_id" : ObjectId("4c47b3372a9b2be866da226e"), "name" : "user0", "age" : 0 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da226f"), "name" : "user1", "age" : 1 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2270"), "name" : "user2", "age" : 2 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2276"), "name" : "user8", "age" : 8 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2277"), "name" : "user9", "age" : 9 } 

> db.users.find({$where: function(){ return this.age > 7 || this.age < 3;}}) 
{ "_id" : ObjectId("4c47b3372a9b2be866da226e"), "name" : "user0", "age" : 0 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da226f"), "name" : "user1", "age" : 1 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2270"), "name" : "user2", "age" : 2 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2276"), "name" : "user8", "age" : 8 } 
{ "_id" : ObjectId("4c47b3372a9b2be866da2277"), "name" : "user9", "age" : 9 }
>>> for u in db.users.find({"$where":"this.age > 7 || this.age < 3"}): print u 
... 
{u'age': 0.0, u'_id': ObjectId('4c47b3372a9b2be866da226e'), u'name': u'user0'} 
{u'age': 1.0, u'_id': ObjectId('4c47b3372a9b2be866da226f'), u'name': u'user1'} 
{u'age': 2.0, u'_id': ObjectId('4c47b3372a9b2be866da2270'), u'name': u'user2'} 
{u'age': 8.0, u'_id': ObjectId('4c47b3372a9b2be866da2276'), u'name': u'user8'} 
{u'age': 9.0, u'_id': ObjectId('4c47b3372a9b2be866da2277'), u'name': u'user9'} 

>>> for u in db.users.find().where("this.age > 7 || this.age < 3"): print u 
... 
{u'age': 0.0, u'_id': ObjectId('4c47b3372a9b2be866da226e'), u'name': u'user0'} 
{u'age': 1.0, u'_id': ObjectId('4c47b3372a9b2be866da226f'), u'name': u'user1'} 
{u'age': 2.0, u'_id': ObjectId('4c47b3372a9b2be866da2270'), u'name': u'user2'} 
{u'age': 8.0, u'_id': ObjectId('4c47b3372a9b2be866da2276'), u'name': u'user8'} 
{u'age': 9.0, u'_id': ObjectId('4c47b3372a9b2be866da2277'), u'name': u'user9'} 

>>> for u in db.users.find().where("function() { return this.age > 7 || this.age < 3;}"): print u 
... 
{u'age': 0.0, u'_id': ObjectId('4c47b3372a9b2be866da226e'), u'name': u'user0'} 
{u'age': 1.0, u'_id': ObjectId('4c47b3372a9b2be866da226f'), u'name': u'user1'} 
{u'age': 2.0, u'_id': ObjectId('4c47b3372a9b2be866da2270'), u'name': u'user2'} 
{u'age': 8.0, u'_id': ObjectId('4c47b3372a9b2be866da2276'), u'name': u'user8'} 
{u'age': 9.0, u'_id': ObjectId('4c47b3372a9b2be866da2277'), u'name': u'user9'}