问题
I need to ask you for help with MySQL select query.
Specific example: employees with the spouse and kids.
I have 2 tables already joined into one and now I need to:
1, select the data with grouping them by 'emp' field
2, transform the result with these rules:
- only one row with particular emp (emp-A, emp-B, emp-C)
- each relative (spouse and kids) in succeeding columns (spouse first, kids next)
The table (in fact two joined tables):
+---------+-----------+-----------+------------+
| emp | relation | relative | birthdate |
+---------+-----------+-----------+------------+
| emp-A | spouse | spouse-A | 1970-xx-xx |
| emp-A | kid | kid-A1 | 1971-xx-xx |
| emp-A | kid | kid-A2 | 1972-xx-xx |
| emp-A | kid | kid-A3 | 1973-xx-xx |
| emp-B | spouse | spouse-B | 1980-xx-xx |
| emp-B | kid | kid-B1 | 1981-xx-xx |
| emp-B | kid | kid-B2 | 1982-xx-xx |
| emp-C | kid | kid-C1 | 1991-xx-xx |
| emp-C | kid | kid-C2 | 1992-xx-xx |
+---------+-----------+-----------+------------+
Desired result:
+---------+-----------+-------------+-----------+-------------+-----------+-------------+-----------+-------------+
| emp | spouse | birthdate | kid1 | birthdate1 | kid2 | birthdate2 | kid3 | birthdate3 |
+---------+-----------+-------------+-----------+-------------+-----------+-------------+-----------+-------------+
| emp-A | spouse-A | 1970-xx-xx | kid-A1 | 1971-xx-xx | kid-A2 | 1972-xx-xx | kid-A3 | 1973-xx-xx |
| emp-B | spouse-B | 1980-xx-xx | kid-B1 | 1981-xx-xx | kid-B2 | 1982-xx-xx | | |
| emp-C | | | kid-C1 | 1991-xx-xx | kid-C2 | 1992-xx-xx | | |
+---------+-----------+-------------+-----------+-------------+-----------+-------------+-----------+-------------+
After several hours of unsuccessful searches, I gave up.
I'll be very grateful for some clues is it possible to achieve something like this (one select query would be the best option for me).
Thank you in advance.
Answering to emsoff, the source tables:
employees:
+----+---------+
| id | emp |
+----|---------+
| 1 | emp-A |
| 2 | emp-B |
| 3 | emp-C |
+----|---------+
relatives:
+----+---------+-----------+-----------+------------+
| id | emp_id | relation | relative | birthdate |
+----+---------+-----------+-----------+------------+
| 1 | 1 | spouse | spouse-A | 1970-xx-xx |
| 2 | 1 | kid | kid-A1 | 1971-xx-xx |
| 3 | 1 | kid | kid-A2 | 1972-xx-xx |
| 4 | 1 | kid | kid-A3 | 1973-xx-xx |
| 5 | 2 | spouse | spouse-B | 1980-xx-xx |
| 6 | 2 | kid | kid-B1 | 1981-xx-xx |
| 7 | 2 | kid | kid-B2 | 1982-xx-xx |
| 8 | 3 | kid | kid-C1 | 1991-xx-xx |
| 9 | 3 | kid | kid-C2 | 1992-xx-xx |
+----|---------+-----------+-----------+------------+
Tables joined with employees.id=relatives.emp_id
回答1:
Your data are somewhat difficult to handle.
To maek a pivot table the columsn have to be uniquie, even more, that also five kis could apperar.
It is quite ugly and it works withmyslq 5.7
Schema (MySQL v5.7)
CREATE TABLE employees (
`id` INTEGER,
`emp` VARCHAR(5)
);
INSERT INTO employees
(`id`, `emp`)
VALUES
('1', 'emp-A'),
('2', 'emp-B'),
('3', 'emp-C');
CREATE TABLE relatives (
`id` INTEGER,
`emp_id` INTEGER,
`relation` VARCHAR(6),
`relative` VARCHAR(8),
`birthdate` DATE
);
INSERT INTO relatives
(`id`, `emp_id`, `relation`, `relative`, `birthdate`)
VALUES
('1', '1', 'spouse', 'spouse-A', '1970-01-01'),
('2', '1', 'kid', 'kid-A1', '1971-01-02'),
('3', '1', 'kid', 'kid-A2', '1972-01-01'),
('4', '1', 'kid', 'kid-A3', '1973-01-01'),
('5', '2', 'spouse', 'spouse-B', '1980-02-01'),
('6', '2', 'kid', 'kid-B1', '1981-02-01'),
('7', '2', 'kid', 'kid-B2', '1982-02-01'),
('8', '3', 'kid', 'kid-C1', '1991-03-01'),
('9', '3', 'kid', 'kid-C2', '1992-03-01');
Query #1
SELECT
GROUP_CONCAT(DISTINCT
CONCAT(
'MAX(case when `relation` = "',
`relation`,
'" then `relation` end) AS `',
relation, '`',
',MAX(case when `relation` = "',
`relation`,
'" then `birthdate` end) AS `',
'birthdate_',relation, '`'
)
ORDER BY rownumber
) INTO @sql
FROM
(SELECT
`relation`
,rownumber
FROM
employees e
INNER JOIN
(SELECT
IF(`relation` = 'kid',IF (@empid = `emp_id`,@rn:= @rn+1,@rn:= 1),IF (@empid = `emp_id`,@rn:= @rn,@rn:= 0)) rownumber
,IF(`relation` = 'kid',CONCAT(`relation`,@rn),`relation`) relation
, `relative`
, `birthdate`
,@empid := `emp_id` emp_id
FROM
(SELECT
*
FROM
relatives
ORDER BY `emp_id`,FIELD(`relation`,"spouse","kid"),`birthdate`) rel
,(SELECT @empid := 0) a1
,(SELECT @rn := 0) a2) r ON e.id = r.emp_id) t1
;
SET @sql = CONCAT("SELECT MIN(`emp`), ", @sql, "
FROM (SELECT
`emp_id`,
`emp` ,
`relation`
,rownumber
, `relative`
, `birthdate`
FROM
employees e
INNER JOIN
(SELECT
IF(`relation` = 'kid',IF (@empid = `emp_id`,@rn:= @rn+1,@rn:= 1),IF (@empid = `emp_id`,@rn:= @rn,@rn:= 0)) rownumber
,IF(`relation` = 'kid',CONCAT(`relation`,@rn),`relation`) 'relation'
, `relative`
, `birthdate`
,@empid := `emp_id` 'emp_id'
FROM
(SELECT
*
FROM
relatives
ORDER BY `emp_id`,FIELD(`relation`,'spouse','kid'),`birthdate`) rel
,(SELECT @empid := 0) a1
,(SELECT @rn := 0) a2) r ON e.id = r.emp_id) t1
GROUP BY `emp_id`");
PREPARE stmt FROM @sql;
EXECUTE stmt;
DEALLOCATE PREPARE stmt;
| MIN(`emp`) | spouse | birthdate_spouse | kid1 | birthdate_kid1 | kid2 | birthdate_kid2 | kid3 | birthdate_kid3 |
| ---------- | ------ | ---------------- | ---- | -------------- | ---- | -------------- | ---- | -------------- |
| emp-A | spouse | 1970-01-01 | kid1 | 1971-01-02 | kid2 | 1972-01-01 | kid3 | 1973-01-01 |
| emp-B | spouse | 1980-02-01 | kid1 | 1981-02-01 | kid2 | 1982-02-01 | | |
| emp-C | | | kid1 | 1991-03-01 | kid2 | 1992-03-01 | | |
View on DB Fiddle
来源：https://stackoverflow.com/questions/61413329/mysql-select-group-by-and-transform-rows-to-separate-columns