Highlighted
One Star

deal with null value

Hi all
In the original table in sql server,column flag was 'NULL', when i extract data from this table via talend ,and write it into delimited file , it values '',blank. How can i keep the column flag the same as 'NULL'? Except using tReplace.


Best Regards!
Joe
5 REPLIES
Community Manager

Re: deal with null value

Hi Joe
Can you please test it is null or empty on column flag as 'NULL' after you read from table? For example:
tMssqlInput--main--tJavaRow
on tJavaRow:
if(Input_row.columnName==null){
System.out.println("it is null");
}
or

if(Input_row.columnName.equals(""){
System.out.println("it is empty");
}
Let me know your test result!
Best regards
Shong
----------------------------------------------------------
Talend | Data Agility for Modern Business
One Star

Re: deal with null value

Hi Shong
Yes, I have tested that already. If I write the data into delimited file first,and then write it to database,it is still empty. If I write it into table directly from one database to another,it is ok,with the ?NULL? Value. But,the table has mass data,and if i process it without delimited file ,it may lead to outofmemory exception and low efficency.
Joe
Community Manager

Re: deal with null value

Hi
Don't know why you use a demited file, but you can convert empty to null on tMap after read it from demilted file, eg:
row1.columnName.equals("")?null:row1.columnName
Best regards
Shong
----------------------------------------------------------
Talend | Data Agility for Modern Business
One Star

Re: deal with null value

There isn't a null value for files, just empty. If you want to include the value null you will have to do it in your tMap.
One Star

Re: deal with null value

Thank you shong and janhess
another problem,If the project that i shall do contains a lot of data(more than10,000,000 rows),how can i avoid the outofmemory exception? now we take the way of storing the data into files first ,and then reading data from file on disk . It seems to work fine,except a little slow. Is there any other better ways?
there will be several jobs running parallelly,so we must take care of the memory.

joe