搜索表中的所有列

Question

I want a procedure that will search all columns for non keyboard ascii characters (Dec 16 to Dec 31 or DLE to US) and update the column by replacing them with a space ' ' or nothing '' .

I have a SELECT statement that is finding the rows I need to update, but I have to manually change all columns myself.

SELECT column_name
FROM table_name
WHERE column_name REGEXP '[[.DLE.]-[.US.]]'

Here's the UPDATE script for modifying the column values

UPDATE table
SET
column = replace(column,char(16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31), '')

I want those two to be fused together to a single function or stored procedure but I don't know how, since I'm just starting to learn MySQL.

Answer 1

Disclaimer

Between using REGEXP and CURSOR s to loop through each table and column, these examples are not going to be lightning fast. The speed will obviously vary depending on your environment and I suggest testing them out on development before production

One column in one table

To search a single column on a single table, you basically had the UPDATE as you needed it.

UPDATE t1
SET
column_name = replace(column_name, 
  char(16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31), '')
WHERE column_name REGEXP '[[.DLE.]-[.US.]]'

All columns in one table

To do all columns in a table, you need to identify the table, then loop through the columns using a cursor

DELIMITER $$
CREATE PROCEDURE table_regexp_replace(in_table VARCHAR(128))
BEGIN
  DECLARE done INT DEFAULT FALSE;
  DECLARE search_column VARCHAR(64);

  DECLARE cur1 CURSOR FOR
    SELECT DISTINCT `COLUMN_NAME` FROM `information_schema`.`COLUMNS`
      WHERE `TABLE_NAME` = in_table ORDER BY `ORDINAL_POSITION` ;

  DECLARE CONTINUE HANDLER FOR NOT FOUND SET done = TRUE;

  OPEN cur1;

  read_loop: LOOP
    -- Process the next column
    FETCH cur1 INTO search_column;

    -- If we're done, stop the loop
    IF done THEN
      LEAVE read_loop;
    END IF;

    -- Replace everything in this column matching the regexp
    SET @new_query := CONCAT ('UPDATE ', in_table, 
        ' SET `', search_column, '` = replace(', search_column, 
          ', char(16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31), \'\')
        WHERE ', search_column, ' REGEXP \'[[.DLE.]-[.US.]]\'') ;

    PREPARE stmt FROM @new_query;
    EXECUTE stmt ;
  END LOOP;

  CLOSE cur1;
END$$

DELIMITER ;

Then usage

CALL table_regexp_replace('my_table');

How it works

Looks convoluted, it's actually pretty straight forward.

1. We create a procedure with one parameter in_table which is used to specify the table to work with.
2. Setup a cursor that pulls the column names from the information_schema table, in their correct order
3. Loop through each of those columns, executing the manually created UPDATE statement against each one.

You'll notice anywhere in the UPDATE query that required quotes, they've had to be escaped using \\ .

\'[[.DLE.]-[.US.]]\'

All columns in all tables

You could then use this procedure in a loop through all tables, using a similar method to above. Below is how you'd pull all the table names from information_schema :

SELECT DISTINCT TABLE_NAME FROM information_schema . TABLES WHERE TABLE_SCHEMA = 'your_database_name';