• Mysql子查询、关联查询


    mysql中update、delete、install尽量不要使用子查询

    一、mysql查询的五种子句

            where(条件查询)、having(筛选)、group by(分组)、order by(排序)、limit(限制结果数)
     
           1、where常用运算符:
                比较运算符
                    > ,  < ,=  , != (< >),>=   ,   <= 
                    in(v1,v2..vn) 
                    between v1 and v2    在v1至v2之间(包含v1,v2)
                逻辑运算符
                    not ( ! )  逻辑非
                    or ( || )    逻辑或
                    and ( && )  逻辑与
     
                    where price>=3000 and price <= 5000 or price >=500 and price <=1000
                    取500-1000或者3000-5000的值
                    where price not between 3000 and 5000
                    不在3000与5000之间的值
     
                模糊查询
                    like 像
                    通配符:
                    %  任意字符
                    _   单个字符
                        where goods_name like '诺基亚%'
                        where goods_name like '诺基亚N__'
     
            2、group by 分组
                    一般情况下group需与统计函数(聚合函数)一起使用才有意义
                    如:select goods_id,goods_name,cat_id,max(shop_price) from goods group by cat_id;
                            这里取出来的结果中的good_name是错误的!因为shop_price使用了max函数,那么它是取最大的,而语句中使用了group by 分组,那么goods_name并没有使用聚合函数,它只是cat_id下的第一个商品,并不会因为shop_price改变而改变
                    mysql中的五种统计函数:
                    (1)max:求最大值
                        select max(goods_price) from goods
                          这里会取出最大的价格的值,只有值
                            #查询每个栏目下价格最高的
                            select cat_id,max(goods_price) from goos group by cat_id;
                            #查出价格最高的商品编号
                            select goods_id,max(goods_price) from goods group by goods_id;
                           
                           
                    (2)min:求最小值
                    (3)sum:求总数和
                            #求商品库存总和
                            select sum(goods_number) from goods;
                    (4)avg:求平均值
                            #求每个栏目的商品平均价格
                            select cat_id,avg(goods_price) from goods group by cat_id;
                    (5)count:求总行数
                            #求每个栏目下商品种类
                            select cat_id,count(*) from goods group by cat_id;
     
                       ###要把每个字段名当成变量来理解,它可以进行运算###
                            例:查询本店每个商品价格比市场价低多少;
                            select goods_id,goods_name,goods_price-market_price from goods;
                                查询每个栏目下面积压的货款
                            select cat_id,sum(goods_price*goods_number) from goods group by cat_id;
     
                        ###可以用as来给计算结果取个别名###
                            select cat_id,sum(goods_price * goods_number)  as hk from goods group by cat_id
                            不仅列名可以取别名,表单也可以取别名
     
                3、having 与where 的异同点
     
                        having与where类似,可以筛选数据,where后的表达式怎么写,having后就怎么写
                        where针对表中的列发挥作用,查询数据
                        having对查询结果中的列发挥作用,筛选数据
                        #查询本店商品价格比市场价低多少钱,输出低200元以上的商品
                        select goods_id,good_name,market_price - shop_price as s from goods having s>200 ;
                        //这里不能用where因为s是查询结果,而where只能对表中的字段名筛选
                        如果用where的话则是:
                        select goods_id,goods_name from goods where market_price - shop_price > 200;
     
                        #同时使用where与having
                        select cat_id,goods_name,market_price - shop_price as s from goods where cat_id = 3 having s > 200;
                        #查询积压货款超过2万元的栏目,以及该栏目积压的货款
                        select cat_id,sum(shop_price * goods_number) as t from goods group by cat_id having s > 20000
                        #查询两门及两门以上科目不及格的学生的平均分
                              思路:
                                #先计算所有学生的平均分
                                 select name,avg(score) as pj from stu group by name;
                                #查出所有学生的挂科情况
                                select name,score<60 from stu;
                                        #这里score<60是判断语句,所以结果为真或假,mysql中真为1假为0
                                #查出两门及两门以上不及格的学生
                                select name,sum(score<60) as gk from stu group by name having gk > 1;
                                #综合结果
                                select name,sum(score<60) as gk,avg(score) as pj from stu group by name having gk >1;
                    4、order by
                        (1) order by price  //默认升序排列
                        (2)order by price desc //降序排列
                        (3)order by price asc //升序排列,与默认一样
                        (4)order by rand() //随机排列,效率不高
                            #按栏目号升序排列,每个栏目下的商品价格降序排列
                            select * from goods where cat_id !=2 order by cat_id,price desc;
                    5、limit
                        limit [offset,] N
                        offset 偏移量,可选,不写则相当于limit 0,N
                        N     取出条目
     
                        #取价格第4-6高的商品
                        select good_id,goods_name,goods_price from goods order by good_price desc limit 3,3;
                       
                ###查询每个栏目下最贵的商品
                    思路:
                            #先对每个栏目下的商品价格排序
                            select cat_id,goods_id,goods_name,shop_price from goods order by cat_id,shop_price desc;
                            #上面的查询结果中每个栏目的第一行的商品就是最贵的商品
                            #把上面的查询结果理解为一个临时表[存在于内存中]【子查询】
                            #再从临时表中选出每个栏目最贵的商品
                            select * from (select goods_id,goods_name,cat_id,shop_price from goods order by cat_id,shop_price desc) as t group by cat_id;
                            #这里使用group by cat_id是因为临时表中每个栏目的第一个商品就是最贵的商品,而group by前面没有使用聚合函数,所以默认就取每个分组的第一行数据,这里以cat_id分组
     
                     良好的理解模型:
                        1、where后面的表达式,把表达式放在每一行中,看是否成立
                        2、字段(列),理解为变量,可以进行运算(算术运算和逻辑运算) 
                        3、 取出结果可以理解成一张临时表
      二、mysql子查询
            1、where型子查询
                    (把内层查询结果当作外层查询的比较条件)
                    #不用order by 来查询最新的商品
                    select goods_id,goods_name from goods where goods_id = (select max(goods_id) from goods);
                    #取出每个栏目下最新的产品(goods_id唯一)
                    select cat_id,goods_id,goods_name from goods where goods_id in(select max(goods_id) from goods group by cat_id);
     
            2、from型子查询
                    (把内层的查询结果供外层再次查询)
                    #用子查询查出挂科两门及以上的同学的平均成绩
                        思路:
                            #先查出哪些同学挂科两门以上
                            select name,count(*) as gk from stu where score < 60 having gk >=2;
                            #以上查询结果,我们只要名字就可以了,所以再取一次名字
                            select name from (select name,count(*) as gk from stu having gk >=2) as t;
                            #找出这些同学了,那么再计算他们的平均分
                            select name,avg(score) from stu where name in (select name from (select name,count(*) as gk from stu having gk >=2) as t) group by name;
     
            3、exists型子查询
                    (把外层查询结果拿到内层,看内层的查询是否成立)
                    #查询哪些栏目下有商品,栏目表category,商品表goods
                        select cat_id,cat_name from category where exists(select * from goods where goods.cat_id = category.cat_id);
       三、union的用法
                  (把两次或多次的查询结果合并起来,要求查询的列数一致,推荐查询的对应的列类型一致,可以查询多张表,多次查询语句时如果列名不一样,则取第一次的列名!如果不同的语句中取出的行的每个列的值都一样,那么结果将自动会去重复,如果不想去重复则要加all来声明,即union all)
               ## 现有表a如下
                    id  num
                    a    5
                    b    10
                    c    15
                    d    10
                表b如下
                    id  num
                    b    5
                    c    10
                    d    20
                    e    99
                求两个表中id相同的和
               select id,sum(num) from (select * from ta union select * from tb) as tmp group by id;
                //以上查询结果在本例中的确能正确输出结果,但是,如果把tb中的b的值改为10以查询结果的b的值就是10了,因为ta中的b也是10,所以union后会被过滤掉一个重复的结果,这时就要用union all
                select id,sum(num) from (select * from ta union all select * from tb) as tmp group by id;
                   
                #取第4、5栏目的商品,按栏目升序排列,每个栏目的商品价格降序排列,用union完成
                select goods_id,goods_name,cat_id,shop_price from goods where cat_id=4 union select goods_id,goods_name,cat_id,shop_price from goods where cat_id=5 order by cat_id,shop_price desc;
                【如果子句中有order by 需要用( ) 包起来,但是推荐在最后使用order by,即对最终合并后的结果来排序】
                #取第3、4个栏目,每个栏目价格最高的前3个商品,结果按价格降序排列
                 (select goods_id,goods_name,cat_id,shop_price from goods where cat_id=3 order by shop_price desc limit 3) union  (select goods_id,goods_name,cat_id,shop_price from goods where cat_id=4 order by shop_price desc limit 3) order by shop_price desc;
               
        四、左连接,右连接,内连接
     
                    现有表a有10条数据,表b有8条数据,那么表a与表b的笛尔卡积是多少?
                        select * from ta,tb   //输出结果为8*10=80条
                     
                1、左连接
                   以左表为准,去右表找数据,如果没有匹配的数据,则以null补空位,所以输出结果数>=左表原数据数
     
                    语法:select n1,n2,n3 from ta left join tb on ta.n1= ta.n2 [这里on后面的表达式,不一定为=,也可以>,<等算术、逻辑运算符]【连接完成后,可以当成一张新表来看待,运用where等查询
                     #取出价格最高的五个商品,并显示商品的分类名称
                    select goods_id,goods_name,goods.cat_id,cat_name,shop_price from goods left join category on goods.cat_id = category.cat_id order by  shop_price desc limit 5;       
               2、右连接
                    a left join b 等价于 b right join a
                    推荐使用左连接代替右连接
                    语法:select n1,n2,n3 from ta right join tb on ta.n1= ta.n2
               3、内连接
                    查询结果是左右连接的交集,【即左右连接的结果去除null项后的并集(去除了重复项)】
                    mysql目前还不支持 外连接(即左右连接结果的并集,不去除null项)
                    语法:select n1,n2,n3 from ta inner join tb on ta.n1= ta.n2
            #########
                     例:现有表a
                            name  hot
                             a        12
                             b        10
                             c        15
                        表b:
                            name   hot
                              d        12
                              e        10
                              f         10
                              g        8
                        表a左连接表b,查询hot相同的数据
                        select a.*,b.* from a left join b on a.hot = b.hot
                        查询结果:
                            name  hot   name  hot
                              a       12     d       12
                              b       10     e       10
                              b       10     f        10
                              c       15     null    null
                        从上面可以看出,查询结果表a的列都存在,表b的数据只显示符合条件的项目       
                          再如表b左连接表a,查询hot相同的数据
                            select a.*,b.* from b left join a on a.hot = b.hot
                            查询结果为:
                            name  hot   name  hot
                              d       12     a       12
                              e        10    b       10
                              f        10     b      10
                              g        8     null    null
                        再如表a右连接表b,查询hot相同的数据
                            select a.*,b.* from a right join b on a.hot = b.hot
                            查询结果和上面的b left join a一样
                    ###练习,查询商品的名称,所属分类,所属品牌
                        select goods_id,goods_name,goods.cat_id,goods.brand_id,category.cat_name,brand.brand_name from goods left join category on goods.cat_id = category.cat_id left join brand on goods.brand_id = brand.brand_id limit 5;
  • 相关阅读:
    剑指offer-第五章优化时间和空间效率(从1到n的整数中1出现的次数)
    《需求工程--软件建模与分析》读书笔记05
    《需求工程--软件建模与分析》读书笔记04
    软件需求第二次课后作业
    2018年春季个人阅读计划
    软件需求与分析——大二下需会知识点
    《需求工程--软件建模与分析》读书笔记03
    《需求工程--软件建模与分析》读书笔记02
    《需求工程--软件建模与分析》读书笔记01
    学习过程总结及对教师授课给出的意见和建议
  • 原文地址:https://www.cnblogs.com/xlz307/p/3428200.html
Copyright © 2020-2023  润新知