查询数组
查询数组很容易，对于数组，我们可以这样理解：数组中每一个元素都是这个键值对键的一个有效值，如下面的例子：我们要查询出售apple的水果店:
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
3. { "_id" : ObjectId("502251a309248743250688e1"), "name" : "good fruit", "fruits" : [ "banana", "pear", "orange" ] }
4. { "_id" : ObjectId("502251c109248743250688e2"), "name" : "good fruit", "fruits" : [ "banana", "apple", "tomato" ] }
5. > db.fruitshop.find({"fruits":"apple"});
6. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
7. { "_id" : ObjectId("502251c109248743250688e2"), "name" : "good fruit", "fruits" : [ "banana", "apple", "tomato" ] }
8. >
我们发现只要包含苹果的数组都能被查询出来。如果要通过多个元素来匹配数组，就需要条件操作符"$all"，比如我们要查询既卖apple又卖banana的水果店：
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
3. { "_id" : ObjectId("502251a309248743250688e1"), "name" : "good fruit", "fruits" : [ "banana", "pear", "orange" ] }
4. { "_id" : ObjectId("502251c109248743250688e2"), "name" : "good fruit", "fruits" : [ "banana", "apple", "tomato" ] }
5. > db.fruitshop.find({"fruits":{"$all":["apple","banana"]}});
6. { "_id" : ObjectId("502251c109248743250688e2"), "name" : "good fruit", "fruits" : [ "banana", "apple", "tomato" ] }
7. >
我们看，使用“$all”对数组内元素的顺序没有要求，只要全部包含的数组都能查询出来。数组查询也可以使用精确匹配的方式，即查询条件文档中键值对的值也是数组，如：
1. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
2. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
3. { "_id" : ObjectId("502253c109248743250688e5"), "name" : "good fruit", "fruits" : [ "apple", "orange", "pear", "banana" ] }
4. > db.fruitshop.find({"fruits":["apple","orange","pear"]});
5. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
6. >
如果是精确匹配的方式，MongoDB的处理方式是完全相同的匹配，即顺序与数量都要一致，上述中第一条文档和查询条件的顺序不一致，第三条文档比查询条件文档多一个元素，都没有被匹配成功！
对于数组的匹配，还有一种形式是精确指定数组中某个位置的元素匹配，我们前面提到，数组中的索引可以作为键使用，如我们要匹配水果店售第二种水果是orange 的水果店：
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
3. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
4. { "_id" : ObjectId("502253c109248743250688e5"), "name" : "good fruit", "fruits" : [ "apple", "orange", "pear", "banana" ] }
5. > db.fruitshop.find({"fruits.1":"orange"});
6. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
7. { "_id" : ObjectId("502253c109248743250688e5"), "name" : "good fruit", "fruits" : [ "apple", "orange", "pear", "banana" ] }
8. >
数组索引从0开始，我们匹配第二种水果就用furits.1作为键。
"$size"条件操作符，可以用来查询特定长度的数组的，如我们要查询卖3种水果的水果店：
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
3. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
4. { "_id" : ObjectId("502253c109248743250688e5"), "name" : "good fruit", "fruits" : [ "apple", "orange", "pear", "banana" ] }
5. > db.fruitshop.find({"fruits":{"$size":3}});
6. { "_id" : ObjectId("5022518d09248743250688e0"), "name" : "big fruit", "fruits" : [ "apple", "pear", "orange" ] }
7. { "_id" : ObjectId("5022535109248743250688e4"), "name" : "fruit king", "fruits" : [ "apple", "orange", "pear" ] }
8. >
但条件操作符"$size"不能和其他操作符连用如“$gt”等，这是这个操作符的一个缺陷。使用这个操作符我们只能精确查询某个长度的数组。如果实际中，在查询某个数组时，需要按其长度范围进行查询，这里推荐的做法是：在这个文档中额外增加一个“size”键，专门记录其中数组的大小，在对数组进行"$push"操作同时，将这个“size”键值加1。如下所示：
1. > db.fruitshop.find({"name":"big fruit"});
2. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "apple", "pear", "orange", "strawberry" ], "name" : "big fruit", "size" : 4 }
3. > db.fruitshop.update({"name":"big fruit"},
4. ... {"$push":{"fruits":"banana"}, "$inc":{"size":1}}, false, false);
5. > db.fruitshop.find({"name":"big fruit"});
6. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "apple", "pear", "orange", "strawberry", "banana" ], "name" : "big fruit", "size" : 5 }
7. >
但这个方式和修改器"$addToSet"没法配合使用，因为你无法判断这个元素是否添加到了数组中！
上篇提到了，find函数的第二个参数用于查询返回哪些键，他还可以控制查询返回数组的一个子数组，如下例：我只想查询水果店售卖说过数组的前两个：
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "apple", "pear", "orange", "strawberry", "banana" ], "name" : "big fruit" }
3. { "_id" : ObjectId("5022535109248743250688e4"), "fruits" : [ "apple", "orange", "pear" ], "name" : "fruit king" }
4. { "_id" : ObjectId("502253c109248743250688e5"), "fruits" : [ "apple", "orange", "pear", "banana" ], "name" : "good fruit" }
5. > db.fruitshop.find({}, {"fruits":{"$slice":2}});
6. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "apple", "pear" ], "name" : "big fruit" }
7. { "_id" : ObjectId("5022535109248743250688e4"), "fruits" : [ "apple", "orange" ], "name" : "fruit king" }
8. { "_id" : ObjectId("502253c109248743250688e5"), "fruits" : [ "apple", "orange" ], "name" : "good fruit" }
9. >
“$slice”也可以从后面截取，用复数即可，如-1表明截取最后一个；还可以截取中间部分，如[2,3]，即跳过前两个，截取3个，如果剩余不足3个，就全部返回！
1. > db.fruitshop.find();
2. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "apple", "pear", "orange", "strawberry", "banana" ], "name" : "big fruit" }
3. { "_id" : ObjectId("5022535109248743250688e4"), "fruits" : [ "apple", "orange", "pear" ], "name" : "fruit king" }
4. { "_id" : ObjectId("502253c109248743250688e5"), "fruits" : [ "apple", "orange", "pear", "banana" ], "name" : "good fruit" }
5. > db.fruitshop.find({}, {"fruits":{"$slice":-1}});
6. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "banana" ], "name" : "big fruit" }
7. { "_id" : ObjectId("5022535109248743250688e4"), "fruits" : [ "pear" ], "name" : "fruit king" }
8. { "_id" : ObjectId("502253c109248743250688e5"), "fruits" : [ "banana" ], "name" : "good fruit" }
9. > db.fruitshop.find({}, {"fruits":{"$slice":[3,6]}});
10. { "_id" : ObjectId("5022518d09248743250688e0"), "fruits" : [ "strawberry", "banana" ], "name" : "big fruit" }
11. { "_id" : ObjectId("5022535109248743250688e4"), "fruits" : [ ], "name" : "fruit king" }
12. { "_id" : ObjectId("502253c109248743250688e5"), "fruits" : [ "banana" ], "name" : "good fruit" }
13. >
如果第二个参数中有个键使用了条件操作符"$slice"，则默认查询会返回所有的键，如果此时你要忽略哪些键，可以手动指明！如：
1. > db.fruitshop.find({}, {"fruits":{"$slice":[3,6]}, "name":0, "_id":0});
2. { "fruits" : [ "strawberry", "banana" ] }
3. { "fruits" : [ ] }
4. { "fruits" : [ "banana" ] }
5. >
查询内嵌文档
查询文档有两种方式，一种是完全匹查询，另一种是针对键值对查询！内嵌文档的完全匹配查询和数组的完全匹配查询一样，内嵌文档内键值对的数量，顺序都必须一致才会匹配，如下例：
1. > db.staff.find();
2. { "_id" : ObjectId("50225fc909248743250688e6"), "name" : { "first" : "joe", "middle" : "bush", "last" : "Schmoe" }, "age" : 45 }
3. { "_id" : ObjectId("50225fe209248743250688e7"), "name" : { "first" : "joe", "middle" : "bush" }, "age" : 35 }
4. { "_id" : ObjectId("50225fff09248743250688e8"), "name" : { "middle" : "bush", "first" : "joe" }, "age" : 25 }
5. > db.staff.find({"name":{"first":"joe","middle":"bush"}});
6. { "_id" : ObjectId("50225fe209248743250688e7"), "name" : { "first" : "joe", "middle" : "bush" }, "age" : 35 }
7. >
针对内嵌文档特定键值对的查询是最常用的！通过点表示法来精确表示内嵌文档的键：
1. > db.staff.find();
2. { "_id" : ObjectId("50225fc909248743250688e6"), "name" : { "first" : "joe", "middle" : "bush", "last" : "Schmoe" }, "age" : 45 }
3. { "_id" : ObjectId("50225fe209248743250688e7"), "name" : { "first" : "joe", "middle" : "bush" }, "age" : 35 }
4. { "_id" : ObjectId("50225fff09248743250688e8"), "name" : { "middle" : "bush", "first" : "joe" }, "age" : 25 }
5. > db.staff.find({"name.first":"joe", "name.middle":"bush"});
6. { "_id" : ObjectId("50225fc909248743250688e6"), "name" : { "first" : "joe", "middle" : "bush", "last" : "Schmoe" }, "age" : 45 }
7. { "_id" : ObjectId("50225fe209248743250688e7"), "name" : { "first" : "joe", "middle" : "bush" }, "age" : 35 }
8. { "_id" : ObjectId("50225fff09248743250688e8"), "name" : { "middle" : "bush", "first" : "joe" }, "age" : 25 }
9. >
我们看，这样查询，所有有效文档均被查询到了！通过点表示法，可以表示深入到内嵌文档内部的键！利用“点表示法”来查询内嵌文档，这也约束了在插入文档时，任何键都不能包含“.” !!
当内嵌文档变得复杂后，如键的值为内嵌文档的数组，这种内嵌文档的匹配需要一些技巧，如下例：
1. > db.blogs.findOne();
2. {
3. "_id" : ObjectId("502262ab09248743250688ea"),
4. "content" : ".....",
5. "comment" : [
6. {
7. "author" : "joe",
8. "score" : 3,
9. "comment" : "just so so!"
10. },
11. {
12. "author" : "jimmy",
13. "score" : 5,
14. "comment" : "cool! good!"
15. }
16. ]
17. }
18. > db.blogs.find({"comment.author":"joe", "comment.score":{"$gte":5}});
19. { "_id" : ObjectId("502262ab09248743250688ea"), "content" : ".....", "comment" : [ { "author" : "joe", "score" : 3, "comment" : "j
20. ust so so!" }, { "author" : "jimmy", "score" : 5, "comment" : "cool! good!" } ] }
21. >
我们想要查询评论中有叫“joe”并且其给出的分数超过5分的blog文档，但我们利用“点表示法”直接写是有问题的，因为这条文档有两条评论，一条的作者名字叫“joe”但分数只有3，一条作者名字叫“jimmy”，分数却给了5！也就是这条查询条件和数组中不同的文档进行了匹配！这不是我们想要的，我们这里是要使用一组条件而不是单个指明每个键，使用条件操作符“$elemMatch”即可！他能将一组条件限定到数组中单条文档的匹配上：
1. > db.blogs.findOne();
2. {
3. "_id" : ObjectId("502262ab09248743250688ea"),
4. "content" : ".....",
5. "comment" : [
6. {
7. "author" : "joe",
8. "score" : 3,
9. "comment" : "just so so!"
10. },
11. {
12. "author" : "jimmy",
13. "score" : 5,
14. "comment" : "cool! good!"
15. }
16. ]
17. }
18. > db.blogs.find({"comment":{"$elemMatch":{"author":"joe", "score":{"$gte":5}}}});
19. > db.blogs.find({"comment":{"$elemMatch":{"author":"joe", "score":{"$gte":3}}}});
20. { "_id" : ObjectId("502262ab09248743250688ea"), "content" : ".....", "comment" : [ { "author" : "joe", "score" : 3, "comment" : "j
21. ust so so!" }, { "author" : "jimmy", "score" : 5, "comment" : "cool! good!" } ] }
22. >
这样做，结果是正确的！利用条件操作符“$elemMatch”可以组合一组条件，并且还能达到的“点表示法”的模糊查询的效果！
$where
上面提到的所有的键值对的查询方式，我们也可以看出，已经很强大了！但如果实际中真的遇到一种情况无法用上述方式实现时，不用慌，MongoDB为我们提供了终极武器："$where"，用他可以执行任意JavaScript作为查询的一部分！最典型的应用：一个文档，如果有两个键的值相等，就选出来，否则不选：
1. > db.fruitprice.find();
2. { "_id" : ObjectId("50226b4c3becfacce6a22a5b"), "apple" : 10, "banana" : 6, "pear" : 3 }
3. { "_id" : ObjectId("50226ba63becfacce6a22a5c"), "apple" : 10, "watermelon" : 3, "pear" : 3 }
4. > db.fruitprice.find({"$where":function () {
5. ... for(var current in this){
6. ... for(var other in this){
7. ... if(current != other && this[current] == this[other]){
8. ... return true;
9. ... }
10. ... }
11. ... }
12. ... return false;
13. ... }});
14. { "_id" : ObjectId("50226ba63becfacce6a22a5c"), "apple" : 10, "watermelon" : 3, "pear" : 3 }
15. >
我们可以看出，使用"$where"其实就是写了一个javascript函数，MongoDB在查询时，会将每个文档转换成一个javascript对象，然后扔到这个函数中去执行，通过返回结果来判断其是否匹配！在实际使用中，尽量避免使用”$where" 条件操作符，因为其性能很差！在执行过程中，需要把每个档案转化为javascript对象！如果不可避免，则尽量这样写：find({”other“：”......“，......，“$where”:""})，即将"$where"放最后，作为结果调优，让常规查询作为前置过滤条件！这样能减少一些性能损失！
我们这里还可以发现，“$where”条件操作符也是作为外层文档的键使用，昨天说“$or”条件操作符是被作为外层文档的键使用。其余目前遇到的条件操作符都是被作为内层文档的键使用！