Highly qualified SQL writing: compare lines. Don't ask why. Asking is highly qualified..

Environmental preparation

Database version ： MySQL 5.7.20-log

Build table SQL：

Log in and copy

       

        DROP TABLE IF EXISTS `t_ware_sale_statistics`;CREATE TABLE `t_ware_sale_statistics` (  `id` bigint(20) NOT NULL AUTO_INCREMENT COMMENT ' Primary key id',  `business_id` bigint(20) NOT NULL COMMENT ' Business organization code ',  `ware_inside_code` bigint(20) NOT NULL COMMENT ' Commodity self coding ',  `weight_sale_cnt_day` double(16,4) DEFAULT NULL COMMENT ' Average daily sales ',  `last_thirty_days_sales` double(16,4) DEFAULT NULL COMMENT ' lately 30 Daily sales ',  `last_sixty_days_sales` double(16,4) DEFAULT NULL COMMENT ' lately 60 Daily sales ',  `last_ninety_days_sales` double(16,4) DEFAULT NULL COMMENT ' lately 90 Daily sales ',  `same_period_sale_qty_thirty` double(16,4) DEFAULT NULL COMMENT ' In the same period last year 30 Daily sales ',  `same_period_sale_qty_sixty` double(16,4) DEFAULT NULL COMMENT ' In the same period last year 60 Daily sales ',  `same_period_sale_qty_ninety` double(16,4) DEFAULT NULL COMMENT ' In the same period last year 90 Daily sales ',  `create_user` bigint(20) DEFAULT NULL COMMENT ' founder ',  `create_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP COMMENT ' Creation time ',  `modify_user` bigint(20) DEFAULT NULL COMMENT ' Final modifier ',  `modify_time` datetime NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP COMMENT ' Final revision time ',  `is_delete` tinyint(2) DEFAULT '2' COMMENT ' Whether or not to delete ,1： yes ,2： no ',  PRIMARY KEY (`id`) USING BTREE,  KEY `idx_business_ware` (`business_id`,`ware_inside_code`) USING BTREE) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 ROW_FORMAT=DYNAMIC COMMENT=' Commodity sales statistics ';
       
       

        
1.
       

Initialization data

To prepare the 769063 Data

Demand background

Sales of goods under the business organization , The same business organization can sell different goods , The same commodity can be sold in different business institutions , That is to say ： The relationship between business organization and commodity is many to many

Suppose there are now n Organizations , There are several commodities under each organization , How to query the sales of goods in these stores ？

Be specific , It's like this

How to find out 100001 Lower commodity 1000、1001、1003 、 100002 Lower commodity 1003、1004 、 100003 Lower commodity 1006、1008、1009 Sales of

It's like a double-layer list （ A set of goods list in the business organization list ） Query for ; Neither the business organization list nor the commodity list is fixed , It's dynamic

So the problem is ： How to query multiple business organizations , Sales of some goods （ Once I described the problem , Maybe it's more blurred , If only everyone understood the meaning ！）

Circular query

It's easy to think of , Circulate the list of business institutions at the code level , Each business organization checks the database once , The pseudocode is as follows ：

Concrete SQL It's like this

SQL Can go index

Implement a simple , Also understand ,SQL You can also go , Everything seems perfect

But the reality is ： Department development specification constraints , You can't cycle through the database

Oh, Ho , This way can only give up , Find another way

OR Splicing

adopt MyBatis Of dynamic SQL function , Conduct SQL Splicing , It's like this

Concrete SQL It's like this

SQL You can also go

Implement a simple , Also understand ,SQL You can also go , And only query the database once , It seems feasible

The only pity is ： It's a little expensive OR, If there are many business institutions , that SQL It will be longer

As one of the candidates , Let's move on

Mixed query and filtering

It's also using Mybatis Of dynamic SQL , take business_id Put the lists together 、 ware_inside_code Put together , It's like this

Concrete SQL It's like this

SQL You can also go

Implement a simple , Also understand ,SQL You can also go , And only query the database once , It seems to work

however ： The result set found is greater than or equal to the result set we want , Your products , Your delicacies ！

Therefore, you also need to filter the found result set , Filter out the result set we want

Let's be one of the candidates , Let's move on

Line by line comparison

SQL-92 Line to line comparison function is added in , thus , Comparison predicate = 、< 、> and IN The parameters of the predicate are no longer just scalar values , It can also be a list of values

Of course , Still have to use Mybatis Of dynamic SQL , It's like this

Concrete SQL It's like this

SQL Can also walk

Implement a simple ,SQL You can also go , And only query the database once , Feel feasible

It's just ： It's a little hard to understand , Because we usually use so little , So this kind of writing looks strange

in addition , The line comparison is SQL standard , Not a specification for a relational database , In other words, relational databases should support this writing method

summary

1、 Finally, I chose Line by line comparison This way to achieve the requirements , Don't ask me why , Asking is forcing gegao ！

2、 There are many ways to realize a requirement , We need to consider the business and various constraints , Choose the most suitable one

3、 The line comparison is SQL-92 Introduced in ,SQL-92 yes 1992 Specifications formulated in , Line to line comparison is not a new feature , It's a basic function that has existed for a long time ！

​

I think it's good , Don't forget to like it + Forward ！