我有一张 table 1:
id - values
12 - 124,145,135
16 - 254,33,11,456,78
...
使用 SQL,我如何拆分要插入到另一个表中的值以获得:
INSERT INTO table2 (id,cat) VALUES 12, 124;
INSERT INTO table2 (id,cat) VALUES 12, 145;
INSERT INTO table2 (id,cat) VALUES 12, 135;
INSERT INTO table2 (id,cat) VALUES 16, 254;
...
谢谢!
最佳答案
这是在 MySQL 中的实现方式:
INSERT INTO table2
SELECT id,
REPLACE(substring(substring_index(vals, ',', i),
length(substring_index(vals, ',', i - 1)) + 1), ',', '')
FROM (
SELECT a.*, @i := if(@id = a.id, @i + 1, 1) i, @id := a.id
FROM (
SELECT *
FROM table1 a
INNER JOIN information_schema.global_status b ON 1 = 1
ORDER BY a.id
) a
INNER JOIN (SELECT @i := 0, @id := NULL) x
) a
WHERE i <= LENGTH(vals) - LENGTH(REPLACE(vals, ',', '')) + 1
显然,主要问题是将第二列转换为行。
这是它的工作原理(您应该从内部查询开始阅读它):
INSERT INTO table2
SELECT id,
/* Get i-th CSV value from vals (where i ranges between 1
and number of generated rows) */
REPLACE(substring(substring_index(vals, ',', i),
length(substring_index(vals, ',', i - 1)) + 1), ',', '')
FROM (
SELECT a.*,
/* Generate an index for each row. The index values will
go from 1 to n (the number of generated rows) for each row
in table1 */
@i := if(@id = a.id, @i + 1, 1) i,
/* We keep a reference to the previous table1.id, so that
we can reset the counter when we move to the next row
in table1 */
@id := a.id
FROM (
/* Generate sufficient rows
You should have at least: Count(table1.*) x MAX_CSV_VALUES_PER_ROW rows */
SELECT *
FROM table1 a
/* I used information_schema.global_status, but you may use any other table
or even create your own temporary table:
SELECT 1 UNION ALL SELECT 2 ... UNION ALL SELECT n */
INNER JOIN information_schema.global_status b ON 1 = 1
ORDER BY a.id
) a
INNER JOIN (SELECT @i := 0, @id := NULL) x
) a
WHERE
/* Keep only the relevant values; the other values are duplicates.
We're going to keep only those values with indices that are less
or equal to the number of CSV values in vals */
i <= LENGTH(vals) - LENGTH(REPLACE(vals, ',', '')) + 1;
关于Mysql 拆分行到 INSERT,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/36649752/