Sql 合并和组？

好的。我有一个如下的查询：

SELECT
    SUM(`order_items`.`quantity`) as `count`,
    `menu_items`.`name`
FROM 
    `orders`,
    `menu_items`,
    `order_items` 
WHERE 
    `orders`.`id` = `order_items`.`order_id` AND 
    `menu_items`.`id` = `order_items`.`menu_item_id` AND 
    `orders`.`date` >= '2008-11-01' AND 
    `orders`.`date` <= '2008-11-30' 
GROUP BY 
    `menu_items`.`id`

选择
总和（`order\u items`.`quantity`）为`count`，
`菜单项`.`名称`
从…起
`订单`，
`菜单项`，
`订购商品`
哪里
`订单`.`id`=`order\u items`.`order\u id`和
`菜单项`.`id`=`订单项`.`菜单项`.`id`和
`订单`.`日期`>='2008-11-01'和
`orders`.`date`Randy的答案很接近，但是where语句删除了对那些不属于该日期范围内任何订单的项目的任何提及

请注意，“左连接”与where子句中链接表的方式不同（即内部连接）。我建议您仔细阅读不同类型的SQL联接（内部联接、外部联接、交叉联接）

在essense中，您需要将从Randy的查询中获得的数据与您的源项列表相关联。使用子选择将执行以下操作：

SELECT
    name
    , nvl(count, 0) as count
FROM 
    menu_items items 
    LEFT JOIN (
        SELECT
            menu_items.id
            , SUM(order_items.quantity) as count
        FROM 
            menu_items
            LEFT JOIN order_items ON menu_items.id = order_items.menu_item_id
            LEFT JOIN orders ON orders.id = order_items.order_id
        WHERE
            "date" between to_date('2008-11-01','YYYY-MM-DD') and to_date('2008-11-30','YYYY-MM-DD')
        GROUP BY
            menu_items.id
    ) counts on items.id = counts.id;

这是在Oracle 10g BTW中。我怀疑您正在使用Oracle，所以您需要转换到您自己的数据库

运行测试将显示以下内容：

SQL> create table menu_items ( id number, name varchar2(10));
create table order_items (order_id number, menu_item_id number, quantity number);
create table orders (id number, "date" date);

Table created.

SQL> 
Table created.

SQL> 
Table created.

SQL> 
insert into menu_items values (1, 'bread');
insert into menu_items values (2, 'milk');
insert into menu_items values (3, 'honey');
insert into menu_items values (4, 'cheese');
SQL> 
1 row created.

SQL> 
1 row created.

SQL> 
1 row created.

SQL> 
1 row created.

SQL> 
insert into orders values (1, to_date('2008-11-02', 'YYYY-MM-DD'));
insert into orders values (2, to_date('2008-11-03', 'YYYY-MM-DD'));
insert into orders values (3, to_date('2008-10-29', 'YYYY-MM-DD'));SQL> 
1 row created.

SQL> 
1 row created.

SQL> 
insert into order_items values (1, 1, 1);
insert into order_items values (1, 3, 1);
1 row created.

SQL> 
1 row created.

SQL> 
insert into order_items values (2, 1, 1);
insert into order_items values (2, 2, 1);
insert into order_items values (2, 3, 1);

insert into order_items values (3, 4, 10);
1 row created.

SQL> 
1 row created.

SQL> 
1 row created.

SQL> 
1 row created.

SQL> SQL> 

1 row created.

SQL> 
SELECT
    name
    , nvl(count, 0) as count
FROM 
    menu_items items 
    LEFT JOIN (
        SELECT
            menu_items.id
            , SUM(order_items.quantity) as count
        FROM 
            menu_items
            LEFT JOIN order_items ON menu_items.id = order_items.menu_item_id
            LEFT JOIN orders ON orders.id = order_items.order_id
        WHERE
            "date" between to_date('2008-11-01','YYYY-MM-DD') and to_date('2008-11-30','YYYY-MM-DD')
        GROUP BY
            menu_iteSQL>   2    3    4    5    6    7  ms.id
    ) counts on items.id = counts.id;  8    9   10   11   12   13   14   15   16   17   18  

NAME            COUNT
---------- ----------
bread               2
milk                1
honey               2
cheese              0

SQL> 
drop table menu_items;
drop table order_items;
drop table orders;SQL> 
Table dropped.

SQL> 
Table dropped.

SQL> 

Table dropped.

SQL> 

PS：使用“date”作为列名是一种不好的做法，因为它（在大多数情况下）是一种类型名，并且可能会导致查询和解析问题。
如果将日期条件放在
JOIN
子句中，这可以在没有任何子查询的情况下完成

下面是我在MySQL 5.0上测试的代码

SELECT m.name, COALESCE(SUM(oi.quantity), 0) AS count
FROM menu_items AS m
  LEFT OUTER JOIN (
    order_items AS oi JOIN orders AS o
      ON (o.id = oi.order_id)
  ) ON (m.id = oi.menu_item_id
      AND o.`date` BETWEEN '2008-11-01' AND '2008-11-30')
GROUP BY m.id;

输出：

+--------+-------+
| name   | count |
+--------+-------+
| bread  |     2 | 
| milk   |     1 | 
| honey  |     2 | 
| cheese |     0 | 
+--------+-------+

以下是MySQL风格的DDL和设置代码：

DROP TABLE IF EXISTS menu_items;
CREATE TABLE menu_items (
  id            INT PRIMARY KEY,
  name          VARCHAR(10)
) TYPE=InnoDB;

DROP TABLE IF EXISTS orders;
CREATE TABLE orders (
  id            INT PRIMARY KEY,
  `date`        DATE
) TYPE=InnoDB;

DROP TABLE IF EXISTS order_items;
CREATE TABLE order_items (
  order_id      INT,
  menu_item_id  INT,
  quantity      INT,
  PRIMARY KEY (order_id, menu_item_id),
  FOREIGN KEY (order_id) REFERENCES orders(id),
  FOREIGN KEY (menu_item_id) REFERENCES menu_items(id)
) TYPE=InnoDB;

INSERT INTO menu_items VALUES
  (1, 'bread'),
  (2, 'milk'),
  (3, 'honey'),
  (4, 'cheese');

INSERT INTO orders VALUES
  (1, '2008-11-02'),
  (2, '2008-11-03'),
  (3, '2008-10-29');

INSERT INTO order_items VALUES
  (1, 1, 1),
  (1, 3, 1),
  (2, 1, 1),
  (2, 2, 1),
  (2, 3, 1),
  (3, 4, 10);

做得好，杰米。1子查询比我的少，而且更合理。MySQL唯一需要改变的是去掉Oracle的日期函数+1我的需求显然比我发布的查询复杂一点，但我用它来构建，现在它运行得很好。谢谢。