Mysql查询动态转换行到列

Question

Can MySQL convert columns into rows, dynamically adding as many columns as are needed for the rows. I think my question might be related to pivot tables but I'm unsure and I don't know how to frame this question other than by giving the following example.

Given a two tables A and B, which look like

Table A

+--+-----+----+
|id|order|data|
+--+-----+----+
|1 |1    |P   |
+--+-----+----+
|2 |2    |Q   |
+--+-----+----+
|2 |1    |R   |
+--+-----+----+
|1 |2    |S   |
+--+-----+----+

I like to write a query that looks like the following:

Result Table

+--+-----+-----+
|id|data1|data2|
+--+-----+-----+
|1 |P    |S    |
+--+-----+-----+
|2 |R    |Q    |
+--+-----+-----+

Basically I want to turn each row in table B into a column in the result table. If there was a new entry was added to table B for id=1, then I want the result table to automatically extend by one column to accommodate this extra data point.

Answer 1

You can use GROUP BY and MAX to simulate pivot. MySQL also supports IF statement.

SELECT  ID,
        MAX(IF(`order` = 1, data, NULL)) data1,
        MAX(IF(`order` = 2, data, NULL)) data2
FROM    TableA
GROUP   BY ID

• SQLFiddle Demo

If you have multiple values of order , dynamic SQL may be more appropriate so that you will not have to modify the query:

SET @sql = NULL;
SELECT
  GROUP_CONCAT(DISTINCT
    CONCAT(
      'MAX(IF(`order` = ', `order`, ',data,NULL)) AS data', `order`)
  ) INTO @sql
FROM TableName;

SET @sql = CONCAT('SELECT  ID, ', @sql, ' 
                  FROM    TableName
                  GROUP   BY ID');

PREPARE stmt FROM @sql;
EXECUTE stmt;
DEALLOCATE PREPARE stmt;

• SQLFiddle Demo
• SQLFiddle Demo ( Another example )

OUTPUT OF BOTH QUERIES:

╔════╦═══════╦═══════╗
║ ID ║ DATA1 ║ DATA2 ║
╠════╬═══════╬═══════╣
║  1 ║ P     ║ S     ║
║  2 ║ R     ║ Q     ║
╚════╩═══════╩═══════╝

Answer 2

You need to use MAX and GROUP BY to simulate a PIVOT:

SELECT Id,
   MAX(CASE WHEN Order = 1 THEN data END) data1,
   MAX(CASE WHEN Order = 2 THEN data END) data2
FROM TableA
GROUP BY Id

And here is the SQL Fiddle .