SQL套题

我一直觉得这种SQL套题是自己明显的短板，之前海康的笔试也遇到了，这也可能是没能通过的一大原因
因为好像到目前为止，除了上《数据库原理》课的时候，其他时候基本都是写的非常简单的增删查改SQL

正文

题目如下：

建表，准备素材

这里省去了goods表的detail字段，同时也省去了外键
先填type表字段

SET NAMES utf8mb4;
SET FOREIGN_KEY_CHECKS = 0;
 
-- ----------------------------
-- Table structure for goods
-- ----------------------------
DROP TABLE IF EXISTS `goods`;
CREATE TABLE `goods`  (
  `GoodsLd` int NOT NULL,
  `Name` varchar(50) CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL,
  `TypeId` int NULL DEFAULT NULL,
  `Stock` int NULL DEFAULT NULL,
  PRIMARY KEY (`GoodsLd`) USING BTREE
) ENGINE = InnoDB CHARACTER SET = utf8 COLLATE = utf8_general_ci ROW_FORMAT = Dynamic;
 
-- ----------------------------
-- Records of goods
-- ----------------------------
INSERT INTO `goods` VALUES (1, '大众S1', 3, 67);
INSERT INTO `goods` VALUES (2, '大众A7', 3, 32);
INSERT INTO `goods` VALUES (3, '丰田U2', 4, 12);
INSERT INTO `goods` VALUES (4, '铃木4', 5, 46);
INSERT INTO `goods` VALUES (5, '宝马Q4', 6, 43);
 
SET FOREIGN_KEY_CHECKS = 1;
 
-- ----------------------------
-- Table structure for order
-- ----------------------------
DROP TABLE IF EXISTS `order`;
CREATE TABLE `order`  (
  `OrderId` int NOT NULL,
  `GoodsId` int NULL DEFAULT NULL,
  `Num` int NULL DEFAULT NULL,
  `Date` datetime NULL DEFAULT NULL,
  PRIMARY KEY (`OrderId`) USING BTREE
) ENGINE = InnoDB CHARACTER SET = utf8 COLLATE = utf8_general_ci ROW_FORMAT = Dynamic;
 
-- ----------------------------
-- Records of order
-- ----------------------------
INSERT INTO `order` VALUES (1, 1, 13, NULL);
INSERT INTO `order` VALUES (2, 3, 4, NULL);
INSERT INTO `order` VALUES (3, 4, 10, NULL);
INSERT INTO `order` VALUES (4, 5, 36, NULL);
 
SET FOREIGN_KEY_CHECKS = 1;
 
-- ----------------------------
-- Table structure for type
-- ----------------------------
DROP TABLE IF EXISTS `type`;
CREATE TABLE `type`  (
  `TypeId` int NOT NULL,
  `Name` varchar(50) CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL,
  `UpId` int NULL DEFAULT NULL,
  PRIMARY KEY (`TypeId`) USING BTREE
) ENGINE = InnoDB CHARACTER SET = utf8 COLLATE = utf8_general_ci ROW_FORMAT = Dynamic;
 
-- ----------------------------
-- Records of type
-- ----------------------------
INSERT INTO `type` VALUES (1, '汽车', NULL);
INSERT INTO `type` VALUES (2, '摩托车', NULL);
INSERT INTO `type` VALUES (3, '大众', 1);
INSERT INTO `type` VALUES (4, '丰田', 1);
INSERT INTO `type` VALUES (5, '铃木', 2);
INSERT INTO `type` VALUES (6, '宝马', 2);
 
SET FOREIGN_KEY_CHECKS = 1;

开始干正事儿

1. 第一问的难点在于上级类别

SELECT
	a.`Name`,
	Stock,
	`类别`,
	c.`name` AS '上级类别' 
FROM
	(
	SELECT
		goods.NAME,
		Stock,
		type.`Name` AS '类别',
		UpId 
	FROM
		goods,
		type 
	WHERE
		Stock > 0 
		AND goods.`Name` LIKE '新品%' 
		AND goods.TypeId = type.TypeId 
	) a
	LEFT JOIN ( SELECT TypeId, NAME FROM type ) c ON c.TypeId = a.UpId

我们先做了一次查询，得到了需要的大部分数据，但是UpId并不是最终的数据，还得再去type表中查一次，而且是需要同时匹配多个数据，同时需要左表中所有的数据，所以这里用LEFT JOIN
另外UpId和TypeId这两个额外的关联字段是必须的

# 还可以先处理type表
SELECT
	goods.`Name`,
	Stock,
	`类别`,
	`上级类别` 
FROM
	goods,(
	SELECT
		a.`TypeId`,
		a.`Name` AS '类别',
		b.`Name` AS '上级类别' 
	FROM
		type a
		LEFT JOIN type b ON a.UpId = b.TypeId 
	) c 
WHERE
	Stock > 0 
	AND goods.`Name` LIKE '新品%' 
	AND goods.TypeId = c.TypeId

2. 这里注意出现的问题是
   这里
   

表1Stock有值，但是表2不存在对应项，导致

这里相减成了一个数减null，结果是null

表1新品化妆品是有值的，但是表2没有这个记录

这里要用CASE WHEN配合ISNULL()解决

SELECT
	`Name`,
	( CASE WHEN ISNULL( sum ) THEN Stock ELSE Stock - sum END ) AS '剩余库存' 
FROM
	(
	SELECT
		goods.`Name`,
		Stock,
		goods.GoodsId 
	FROM
		goods,
		type 
	WHERE
		goods.TypeId = type.TypeId 
	AND goods.TypeId = ( SELECT TypeId FROM type WHERE NAME = '化妆品' )) a
	LEFT JOIN ( SELECT GoodsId, SUM( Num ) AS sum FROM `order` GROUP BY GoodsId ) b ON a.GoodsId = b.GoodsId

3. 第三题的问题在于，查询前三，但是要求前三的数量可能不止3个

这里是涉及了排名函数

# 这么写的问题是查不出并列的情况
SELECT
	`销售数量`,
	b.`Name` 
FROM
	( SELECT GoodsId, SUM( Num ) AS '销售数量' FROM `order` GROUP BY GoodsId ORDER BY `销售数量` DESC LIMIT 3 ) a
	LEFT JOIN goods b ON a.GoodsId = b.GoodsId

窗口函数

涉及到了不会的窗口函数，学习一下

这是我的原始数据

SELECT *,
RANK() over (ORDER BY 成绩 desc) as 'rank',
DENSE_RANK() over (ORDER BY 成绩 desc) as 'dense_rank',
ROW_NUMBER() over (ORDER BY 成绩 desc) as 'row_num'
from class;

结果

注意这里

可以看到，能够相同成绩同一排名的是RANK()和DENSE_RANK()函数，区别是：

• RANK()尽管排名相同，但是会占用后面的排名
• DENSE_RANK()排名相同，不占用后面的排名，接着排

那么如果考虑要分组呢？
添加partition by

select *,
   rank() over (partition by 班级
                 order by 成绩 desc) as ranking
from 班级表

然后再回看这道题
先查出id和id对应的所有订单的总销量
原本是这么写的

SELECT GoodsId, SUM( Num ) AS '销售数量' FROM `order` GROUP BY GoodsId ORDER BY `销售数量` DESC

ORDER BY都可以改写，但是GROUP BY要保留
这里应该是RANK()和DENSE_RANK()都可以

这样就排出来了

SELECT
	GoodsId,
	`销售数量` 
FROM
	(
	SELECT
		GoodsId,
		`销售数量`,
		RANK() over ( ORDER BY `销售数量` DESC ) AS '排名' 
	FROM
		( SELECT GoodsId, SUM( Num ) AS '销售数量' FROM `order` GROUP BY GoodsId ) c 
	) a 
WHERE
	`排名` <4

看起来可以说很不优雅了