MySql-通过将行转置为列来生成报告的Sql查询或过程

Question

This is the first time - I'm trying to do something like this - so please bear with me. This is on MySql.

I am trying to generate a report to see which students have completed which topics and on which dates.

This is the current query that I run

select u.email,t.topic_name,tu.date_created as 'date completed'
from topic_user tu
join topic t ON tu.topic_id = t.topic_id
join user u ON tu.user_id = u.user_id

which will return results like

email         |    topic_name      |    date completed
abc@gmail.com |    ABC             |    03/01/2012
abc@gmail.com |    DEF             |    03/02/2012
abc@gmail.com |    ABC             |    03/08/2012
abc@gmail.com |    GHI             |    03/08/2012
def@gmail.com |    ABC             |    03/02/2012
def@gmail.com |    XYZ             |    03/10/2012

The way I want to generate the report is have the topic names as column headers and the date they completed it as values

email         |    ABC      |    DEF      |    GHI      |    JKL      |    XYZ     
abc@gmail.com | 03/08/2012  | 03/02/2012  | 03/08/2012  |    null     |    null
def@gmail.com | 03/02/2012  |    null     |    null     |    null     |  03/10/2012

Few things to note are:

1) All the topic names would come from the topic table - even if they have not been completed by the students - the value should appear as null

2) Incase of student abc@gmail.com - he has studied the topic ABC twice - but the report should get the latest date.

I guess I have to write a stored procedure to accomplish this. Like maybe first pull all the topic names from the topic table and then create a temp view and populate it.

I would appreciate any help you can provide. Thanks much

Answer 1

I've not tested this, and my experience with MySQL is limited but I hope the below is what you are after. It dynamically creates the SELECT statement using the GROUP_CONCAT function, then executes it (This is the bit I am not certain of the way to do it in MySQL).

SET @SQL = (
SELECT  CONCAT('SELECT Email,', GROUP_CONCAT(SelectText), ' FROM Topic_User tu INNER JOIN Users u ON u.User_ID = tu.User_ID GROUP BY Email')
FROM    (   SELECT  CONCAT(' MAX(CASE WHEN Topic_ID = ', Topic_ID, ' THEN tu.Date_Created END) AS `', Topic_Name, '`') AS SelectText
            FROM    Topic
        )  AS d);

PREPARE stmt FROM @SQL;
EXECUTE stmt

Of course if your topics are not changing very regularly you could just use:

SELECT  Email,
        MAX(CASE WHEN Topic_ID = 1 THEN tu.Date_Created END) AS ABC,
        MAX(CASE WHEN Topic_ID = 2 THEN tu.Date_Created END) AS DEF,
        MAX(CASE WHEN Topic_ID = 3 THEN tu.Date_Created END) AS GHI,
        MAX(CASE WHEN Topic_ID = 4 THEN tu.Date_Created END) AS JKL,
        MAX(CASE WHEN Topic_ID = 5 THEN tu.Date_Created END) AS XYZ
FROM    Topic_User tu
        INNER JOIN users u
            ON u.User_ID = tu.User_ID
GROUP BY Email

and alter the query each time a new topic is added (This is the query produced by the process above).

Answer 2

You could do this with a dynamically generated crosstab query. The query you want to end up with is something like this -

SELECT
    u.email,
    MAX(IF(t.topic_name = 'ABC', tu.date_created, NULL)) AS 'ABC',
    MAX(IF(t.topic_name = 'DEF', tu.date_created, NULL)) AS 'DEF',
    etc
FROM topic_user tu
INNER JOIN topic t
    ON tu.topic_id = t.topic_id
INNER JOIN user u
    ON tu.user_id = u.user_id
GROUP BY u.user_id;

So, in your server side language of choice you would dynamically build your field list based on first querying for a list of topics.

You could slightly reduce the overhead of this query by using the topic_ids instead of the topic_names so that you can drop the join to the topics table -

SELECT
    u.email,
    MAX(IF(tu.topic_id = 1, tu.date_created, NULL)) AS 'ABC',
    MAX(IF(tu.topic_id = 2, tu.date_created, NULL)) AS 'DEF',
    etc
FROM topic_user tu
INNER JOIN user u
    ON tu.user_id = u.user_id
GROUP BY u.user_id;