Reading or writing the last record to a Sequential File

[durgaps]

Hi,

I have a source text file which has 10 recs and I am having a few transformations in the middle and putting the output to a sequential file again having 10 recs. Is there any way I can write only the last record i.e. the 10th record to the file.

Thanks,

[ray.wurlod]

Aggregator stage. Set the aggregation function to Last for every column.

[prabu]

Hi,

I have a source text file which has 10 recs and I am having a few transformations in the middle and putting the output to a sequential file again having 10 recs. Is there any way I can write only the last record i.e. the 10th record to the file.

Thanks,

check for @OUTROWNUM and mod function. But please beware, when you use parallel jobs, then your definition of 10th record is no more valid.

example:-
Select the 10th record from a given table.
10th record can be the physical location, the 10th record from your data buffer, or may be the 10th record after sorting by date

you may also want to try @PARTITIONNUM and @NUMPARTITIONS to define your 10th record

[durgaps]

Hi,

I have a source text file which has 10 recs and I am having a few transformations in the middle and putting the output to a sequential file again having 10 recs. Is there any way I can write only the last record i.e. the 10th record to the file.

Thanks,

check for @OUTROWNUM and mod function. But please beware, when you use parallel jobs, then your definition of 10th record is no more valid.

example:-
Select the 10th record from a given table.
10th record can be the physical location, the 10th record from your data buffer, or may be the 10th record after sorting by date

you may also want to try @PARTITIONNUM and @NUMPARTITIONS to define your 10th record

Hi Ray,

Thanks for the reply. I am using parallel extender and I dont find the LAST function in AGGREGATOR stage

[m_keerthi2005]

The last option may also be available in PX. when you use group by on any column then you may have to use last option.

If you are doing any group by, then other option could be use the MAX(Column) in the derivation column. this should solve your problem.

[keshav0307]

you can have a Trail stage to get the last N records.

[kumar_s]

@OUTROWNUM can be used only if the number of records are predefined. Tail stage can be possible used based on your requirement.

[talk2shaanc]

if you want to read the last rows from the source file, you can use "external source" stage and then use unix command to get the last row.
tail -1 /#path#/filename.dat