Reputation: 557
i have to sum element in a column(SCENARIO1) that is varchar and contain data like (1,920, 270.00, 0, NULL) but when i try to convert data into int or decimal i get this error : "he command is wrong when converting value "4582014,00" to int type"
here is my request :
select sum( convert( int, SCENARIO1) )
from Mnt_Scenario_Exercice where code_pere='00000000'
any help please
Upvotes: 2
Views: 21567
Reputation: 18940
The problem is due to the fact that the sum function isn't decoding SCENARIO1 as containing a CSV list of numbers. The way the sum function is usually used is to sum a lot of numbers drawn from multiple rows, where each row provides one number.
Try doing it in two steps. In step 1 convert the table into first normal form perhaps by UNPIVOTING. The 1NF table will have one number per row, and will contain more rows than the initial table.
The second step is to compute the sum. If you want more than one sum in the result, use GROUP BY to create groups, and then select a sum(somecolumn). This will yield one sum for each group.
Upvotes: 1
Reputation: 18759
Try this, I haven't got a way to test yet, but I will test and replace if incorrect.
SELECT sum(CAST (replace(SCENARIO1, ',', '') AS INT))
FROM Mnt_Scenario_Exercice
WHERE code_pere = '00000000';
EDIT: You can use a numeric
for the cast if you need 4582014,00 to be 4582014.00
SELECT sum(CAST (replace(SCENARIO1, ',', '.') AS NUMERIC(10,2)))
FROM Mnt_Scenario_Exercice
WHERE code_pere = '00000000';
Upvotes: 0
Reputation: 1526
4582014,00 should be a decimal
try this (I assume that youre query is working) and changed convert(int into decimal)
select sum(convert(decimal(20,2),replace(SCENARIO1, ',', '.'))) from Mnt_Scenario_Exercice where code_pere='00000000'
Upvotes: 2
Reputation: 117380
try this
select sum(cast(replace(SCENARIO1, ',', '.') as decimal(29, 10)))
from Mnt_Scenario_Exercice
where code_pere = '00000000';
If you couldn't convert your '4582014,00'
into decimal, there's a chance you have different decimal separator on your server. You could look what it is or just try '.'
Upvotes: 3