动态SQL：为具有关系的表创建插入查询

Question

我有2个表，foo和bar。 bar持有foo的外键，例如：

foo

id | some_data
-: | :--------
 1 | a
 2 | b
 3 | c

bar

id | id_foo | more_data
-: | -----: | :--------
 1 |      1 | d
 2 |      1 | e
 3 |      2 | f
 4 |      3 | g
 5 |      3 | h
 6 |      3 | i

我想动态创建查询以插入到foo和bar的其他位置。

对于表foo，这很容易做到：

SELECT CONCAT('INSERT INTO new_foo (some_data) VALUES (''', f.some_data, ''');') AS 'sql'
FROM foo f;

我得到了预期的结果：

INSERT INTO new_foo (some_data) VALUES ('a');
INSERT INTO new_foo (some_data) VALUES ('b');
INSERT INTO new_foo (some_data) VALUES ('c');

但是，我坚持为包含外键的表生成插入查询。

我以为使用2个UNION，一个获取最后插入的ID，第二个使用该ID生成对bar的查询：

SELECT sql
FROM
(
    SELECT f.id AS 'id', CONCAT('INSERT INTO new_foo (some_data) VALUES (''', f.some_data, ''');') AS 'sql'
    FROM foo f
    UNION
    SELECT f.id AS 'id', 'SET @last_id = SELECT LAST_INSERT_ID();' AS 'sql' -- Target will be MySQL
    FROM foo f
    UNION
    SELECT b.id_foo AS 'id', CONCAT('INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, ''', b.more_data, ''');') AS 'sql'
    FROM bar b
) s
ORDER BY s.id;

这给了我输出：

INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'd');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'e');
INSERT INTO new_foo (some_data) VALUES ('a');
SET @last_id = SELECT LAST_INSERT_ID();
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'f');
INSERT INTO new_foo (some_data) VALUES ('b');
SET @last_id = SELECT LAST_INSERT_ID();
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'g');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'h');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'i');
INSERT INTO new_foo (some_data) VALUES ('c');
SET @last_id = SELECT LAST_INSERT_ID();

查询顺序错误。它首先生成对bar的查询，然后对foo，然后对LAST_INSERT_ID()进行查询。

应该是：

INSERT INTO new_foo (some_data) VALUES ('a');
SET @last_id = SELECT LAST_INSERT_ID();
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'd');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'e');
INSERT INTO new_foo (some_data) VALUES ('b');
SET @last_id = SELECT LAST_INSERT_ID();
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'f');
INSERT INTO new_foo (some_data) VALUES ('c');
SET @last_id = SELECT LAST_INSERT_ID();
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'g');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'h');
INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, 'i');

如何确保遵守正确的顺序？我可以处理每个查询在UNION中的位置，但是不确定是否将保留条目的顺序。

这里是fiddle要玩

Answer 1

您有三行具有相同的ID，并且未指定顺序。您只需添加新列即可指定顺序：

SELECT sql
FROM
(
    SELECT f.id AS 'id', -1 AS seq, CONCAT('INSERT INTO new_foo (some_data) VALUES (''', f.some_data, ''');') AS 'sql'
    FROM foo f
    UNION
    SELECT f.id AS 'id', 0 AS seq, 'SET @last_id = SELECT LAST_INSERT_ID();' AS 'sql' -- Target will be MySQL
    FROM foo f
    UNION
    SELECT b.id_foo AS 'id', b.id AS seq, CONCAT('INSERT INTO new_bar (id_foo, more_data) VALUES (@last_id, ''', b.more_data, ''');') AS 'sql'
    FROM bar b
) s
ORDER BY s.id, s.seq;