Update all fields in a table that contain a specific value

Question

I have a MySQL table with 80 fields and I want to update all fields that contain a specified value. For example, I want to update the fields with item1 to the value 0 :

id_num    field1       field2   field3   field4   field5
  1        item1       item2    item3    item4     item5
  2        item3       item3    item3    item4     item5
  3        item2       item1    item3    item4     item5
  4        item5       item4    item3    item2     item1
  5        item1       item1    item3    item4     item5
  6        item2       item2    item4    item4     item5

The result should be:

  id_num    field1     field2   field3   field4   field5
      1        0       item2    item3    item4     item5
      2        item3   item3    item3    item4     item5
      3        item2   0        item3    item4     item5
      4        item5   item4    item3    item2     0
      5        0       0        item3    item4     item5
      6        item2   item2    item4    item4     item5

I want to do it with a MySQL query. The problem I have is that with 80 fields, I would have to use 80 queries like:

update set field1=0 where field1=item1;
update set field2=0 where field2=item1;

And so on, which would be exhausting! :(

Answer 1

In an UPDATE statement, the WHERE clause determines which rows are updated. Here, you also need to determine which column to update; so the only use of the WHERE clause would be to identify and filter out any rows where no columns at all need to change, ie:

UPDATE table
SET
    ...
WHERE field1 = "item1"
    OR field2 = "item1"
    OR field3 = "item1"
    ...;

But judging by the number of columns and what you're trying to do, there's no way you've got a covering index for such a query. And because they're all OR conditions, you're not likely to be able to use even a single very selective index to prevent a full table scan. What that means is that the WHERE clause is probably entirely useless for you here, and your UPDATE is going to be slow on any reasonably large table.

So you can't filter rows, you can only choose which columns to SET ; but in any given row, you might have to set any column. So with SQL, you actually have to SET every column; but apply a condition so that you only change the value when you want to. Use a CASE expression for this:

UPDATE table
SET
    field1 = CASE field1 WHEN "item1" THEN "0" ELSE field1 END,
    field2 = CASE field2 WHEN "item1" THEN "0" ELSE field2 END,
    field3 = CASE field3 WHEN "item1" THEN "0" ELSE field3 END,
    ...;

I'm using "item1" and "0" here just for consistency, since you haven't told us what the datatypes of these columns are. Use your own values, obviously.

Seeing your problem though, the question has to be asked: Why do you have this same "item1" in every column? You should have different sorts of data in each column, not repetitions of the same sort of data over and over again. There appear to be serious design issues with the table you're describing; you should seriously look into a redesign. The fact that you can't filter out any rows with your query is a big warning sign.

Answer 2

In many statements, you can do it like this.

UPDATE table 
SET   field1 = '0' 
WHERE field1 = 'item1';

UPDATE table 
SET   field2 = '0' 
WHERE field2 = 'item1';

...

UPDATE table 
SET   fieldN = '0' 
WHERE fieldN = 'item1';

In one statement, you can do it like this.

UPDATE table 
SET   field1 = CASE WHEN field1 = 'item1'
                 THEN '0' 
                 ELSE field1 
          END
, field2 = CASE WHEN field2 = 'item1'
                 THEN '0' 
                 ELSE field2
          END
...

, fieldN = CASE WHEN fieldN = 'item1'
                 THEN '0' 
                 ELSE fieldN 
          END
;

However, I think this is another wrong question, and no literal answer will do good.

Id you need such a loop, your database design is probably wrong.

A column has to contain distinct value, not shared value among other columns.

Update : (a php solution) This PHP script will go through each column name and apply the query to it.

<?php
$result = mysql_query("SHOW COLUMNS FROM table ");
while ($row = mysql_fetch_assoc($result)) {
    mysql_query("UPDATE table SET ".$row['field']." = '0' WHERE ".$row['field']." = 'item1'");
}
?>

Answer 3

If you have access to a web server with php, you can do this:

<?php

$connection = new mysqli( host, user, password, database );
// Skipping error checking
for ($i = 0; i < num_of_columns; i++) {
  $connection->query( "UPDATE table SET field".i."=0 WHERE field".i."='item1'" );
}
?>

You should also consider database normalization, the act of making data like this manageable.