从同一个表中过滤逗号分隔值

Question

Can anybody help me to get the sub_cat(csv) values which are not in id column from same table.

(sub_cat) comma separated values are the ids of same table, i need to get the values which are not in id column. Like 2,3,7 are present in id column whereas 20,24 are not. I need to get 20,24 only.

Answer 1

As I have elaborated in this post, I recommend not storing data in CSV format. This gives trouble accessing and updating it.

I am not sure about this, but you might be able to simply use:

SELECT sub_cat FROM table_name WHERE id NOT IN (
    SELECT sub_cat FROM table_name
)

However, I always prefer to store only one ID per row.

Answer 2

To do this purely in MySQL required writing a stored procedure; you'll need to change the database name test to the actual name of your database. It would have been a whole lot easier to do it with PHP - but where's the fun in that?

DELIMITER //

CREATE PROCEDURE test.check_subcats(
    IN s_delimiter VARCHAR(30)
    )

    DETERMINISTIC
    READS SQL DATA

    BEGIN
        DECLARE s_csv TEXT;
        DECLARE i_subcat_index INT(10) unsigned DEFAULT 1;
        DECLARE i_subcat_count INT(10) unsigned;
        DECLARE l_category_done INT(10) DEFAULT FALSE;
        DECLARE c_category CURSOR FOR SELECT category.sub_cat FROM category;
        DECLARE CONTINUE HANDLER FOR NOT FOUND SET l_category_done = TRUE;

        -- create a temporary table to hold every csv value
        CREATE TEMPORARY TABLE IF NOT EXISTS tmp_csv( cvalue VARCHAR(10) NOT NULL );

        OPEN c_category;
        l_category: LOOP
            FETCH c_category INTO s_csv;

            IF l_category_done THEN
                LEAVE l_category;
            ELSE
                -- determine the number of sub-categories
                SELECT (LENGTH(s_csv) - LENGTH(REPLACE(s_csv, s_delimiter, ''))) + 1 INTO i_subcat_count;

                -- loop to store all csv values
                WHILE i_subcat_index <= i_subcat_count DO 
                    INSERT INTO tmp_csv ( cvalue ) (
                    SELECT REPLACE(SUBSTRING(
                            SUBSTRING_INDEX(s_csv, s_delimiter, i_subcat_index),
                            LENGTH(SUBSTRING_INDEX(s_csv, s_delimiter, i_subcat_index - 1)) + 1
                        ), s_delimiter, '')
                    );

                    SET i_subcat_index = i_subcat_index + 1;
                END WHILE;
            END IF;

            SET i_subcat_index = 1;
        END LOOP;       
        CLOSE c_category;

        SELECT DISTINCT tmp_csv.cvalue FROM tmp_csv WHERE tmp_csv.cvalue NOT IN ( SELECT category.id FROM category );
        DROP TEMPORARY TABLE IF EXISTS tmp_csv;
    END //

DELIMITER ;

I'm not 100% certain as to how robust it is but it was working with your data on my dev box.

You specify the delimiter for your CSV data when you call the procedure thus:

CALL `check_subcats`(',');

Essentially this loops through the category table to read sub_cat . It then splits the sub_cat value into chunks using the delimiter provided (much like PHPs explode() function) and writes every one of those values to a temporary table.

This then gives you a temporary table holding all your CSV data in individual bits and it's then a simple matter of selecting everything from that data that's NOT IN the category.id list.