測試一下基本的，從查詢結(jié)果里面構(gòu)造一個(gè)json 的格式

create?table?t1(ID?int?identity,name?nvarchar(50),Chinese?int?,Math?int)

insert?into?t1?values?('張三',90,80),('李四',75,90),('王五',68,100)
select?*?from?t1

select?*?from?t1?for?json?auto


--查詢結(jié)果
ID??????????name???????????????????????????????????????????????Chinese?????Math
-----------?--------------------------------------------------?-----------?-----------
1???????????張三?????????????????????????????????????????????????90??????????80
2???????????李四?????????????????????????????????????????????????75??????????90
3???????????王五?????????????????????????????????????????????????68??????????100

--json?格式
[{"ID":1,"name":"張三","Chinese":90,"Math":80},{"ID":2,"name":"李四","Chinese":75,"Math":90},{"ID":3,"name":"王五","Chinese":68,"Math":100}]

?這個(gè)是默認(rèn)模式下面使用json的查詢結(jié)果。是不是十分清晰

然后我們?cè)俳釉賲?，第二波是這樣紙的。假如我們要繼續(xù)搞有層級(jí)關(guān)系的。我們還可以這樣寫。比方說把成績放在一個(gè)叫points 的節(jié)點(diǎn)里面， 也是可以分層的

select?ID,
????????name,
????????Chinese?as?[Points.Chinese],
????????Math?as?[Points.Math]
????from?t1?for?json?path


--結(jié)果json
[
{"ID":1,"name":"張三","Points":{"Chinese":90,"Math":80}},
{"ID":2,"name":"李四","Points":{"Chinese":75,"Math":90}},
{"ID":3,"name":"王五","Points":{"Chinese":68,"Math":100}}
]

他們的分?jǐn)?shù)就放在了json 里面的,被一個(gè)point 包住了。

如果說我要在這個(gè)結(jié)果里面添加一個(gè)頭來包住，當(dāng)然，我可以使用每個(gè)列來個(gè)別名 [root.col] 來實(shí)現(xiàn)，然而就有點(diǎn)啰嗦了。所以我們可以使用這個(gè)root 的關(guān)鍵字來添加一個(gè)頂節(jié)點(diǎn)

select?ID,
????????name,
????????Chinese?as?[Points.Chinese],
????????Math?as?[Points.Math]
????from?t1?for?json?path,root('root')?

--返回的json結(jié)果
{"root":[
????{"ID":1,"name":"張三","Points":{"Chinese":90,"Math":80}},????{"ID":2,"name":"李四","Points":{"Chinese":75,"Math":90}},{"ID":3,"name":"王五","Points":{"Chinese":68,"Math":100}}]}

當(dāng)然咯，查詢嘛，錄入數(shù)據(jù)總是難免遇到null值，在這方面,for json 是如何處理的呢？ 我在測試表添加一條數(shù)據(jù)在來查詢

insert?into?t1?values?('趙六',100,null)


select?ID,
????????name,
????????Chinese?as?[Points.Chinese],
????????Math?as?[Points.Math]
????from?t1?
????????where?id?in(3,?4)
????for?json?auto

--json的返回結(jié)果
[{"ID":3,"name":"王五","Points.Chinese":68,"Points.Math":100},{"ID":4,"name":"趙六","Points.Chinese":100}]

auto 模式下，如果是空值，將會(huì)忽略該屬性。這樣的話很容易就每一個(gè)集合返回的屬性數(shù)量都不一來，這樣不好看。所以應(yīng)對(duì)這種情況，我們可以使用 incluede_null_values 關(guān)鍵字，即使是空值，也帶出來?

select?ID,
????????name,
????????Chinese?as?[Points.Chinese],
????????Math?as?[Points.Math]
????from?t1?
????????where?id?in(3,?4)
????for?json?auto,?include_null_values

--json?的返回結(jié)果
[{"ID":3,"name":"王五","Points.Chinese":68,"Points.Math":100},{"ID":4,"name":"趙六","Points.Chinese":100,"Points.Math":null}]

使用了這個(gè)關(guān)鍵字，就可以把空值帶出來，里面的值是Null 值

?

好，本次實(shí)驗(yàn)到此為止~然后我試下解析json 的語法之類的再分享~

感想就是其實(shí)語法應(yīng)該跟xml類型的相差無幾~但是數(shù)據(jù)庫之前支持了xml 數(shù)據(jù)類型，然后json卻只能通過字符串去轉(zhuǎn)換解析。