SORT Block of records

girishb2 · New User Joined: 30 Sep 2006 Posts: 14

Hi Friends.
I got a new requirement where I need to extract the block of records as per the country. The first two records are header and 3rd record contains country name and then detail records (these vary.. May be 6, may be 10, 12) then at the end I have trailer. This whole becomes a block of record for me. This file is FB with 400 LRECL. I need to recognize these block of records as per the country names and write it into separate files. I believe I need to use IFTHEN, GROUP, BEGIN, END Sort cards to do that but not sure how to. I have attached one block of record here. So I need to extract the whole block/bunch with same sequence for different countries. Please help on sort card.

Pandora-Box · Posted: Thu Sep 15, 2016 11:44 am

Your query is not clear to me atleast
Is there any other reference file where you need to cross check the names of country present in file place as an image above?

Please do not use image use code tags going forward...

mistah kurtz · Posted: Thu Sep 15, 2016 1:57 pm

girishb2 · New User Joined: 30 Sep 2006 Posts: 14

Thanks. I will try to find the posts on this forum.

Hi Pandora, Sorry for not making it clear.
Here is a example - I have a file which contains -

HDR01
hdr
Costa Rica
001
DTL
002
005
TRL
HDR02
hdr
Argentina
001
003
DTL
TRL
HDR01
hdr
Costa Rica
001
TYH
090
005
132
TRL

I need to sort the file with hdr, block of detail records & trailer as per the country names. Like for Costa Rica - I have 2 block of records which I need to extract and keep it in separate files. I need a sort card for this.

Pandora-Box · Posted: Thu Sep 15, 2016 6:00 pm

Argentina is a country name as well why are you not expecting it in the output file?

girishb2 · New User Joined: 30 Sep 2006 Posts: 14

Yes, I need Costa Rica, Argentina, Brazil - block of records in separate files. Like -

file1
HDR01
hdr
Costa Rica
34324
eet
TRL

File2
HDR01
hdr
Argentina
435435
ssafd
dff
TRL

Arun Raj · Posted: Thu Sep 15, 2016 6:23 pm

girishb2,

If you don't know how many country names/blocks are going to be in the input, then you need output files to be dynamically allocated. I would suggest writing some code instead of a sort solution.

sergeyken · Posted: Thu Sep 15, 2016 11:57 pm

1) Your "Costa Rica" appers twice in your input. Do you need to create two separate files for two of them?

2) Is that OK to create library members instead of "files"?

girishb2 · New User Joined: 30 Sep 2006 Posts: 14

No. I need to club these Costa Rica Records into One PS Dataset. Likewise there may be many records for Argetina & Brazil which I need to keep in another PS Dataset.

Rohit Umarjikar · Posted: Fri Sep 16, 2016 1:16 am

DFSORT does NOT allocate the output data sets dynamically. You would either have to supply the maximum number of output DD statements or generate them in a job you submit to the internal reader Or learn the approach explained by Frank here Or a Rexx Solution Or Write a program as suggested by Arun.

girishb2 · New User Joined: 30 Sep 2006 Posts: 14

I already have datasets created. I just want to do the group sort on the basis of Country name and have these in the datasets - Costa Rica in one & Arg, BR in another.

Arun Raj · Posted: Fri Sep 16, 2016 2:16 am

girishb2,

If this is not a one-time task, do you always know how many countries/output datasets you will have? If that is a variable number then dynamic allocation is the way to go.

sergeyken · Posted: Fri Sep 16, 2016 4:20 am

Since the tag to determine the required country does appear only after third record of each group, there is no way to guess in advance: where to place the preceeding "header" records.

There are no such things as miracles...

Some pre-processing is needed to define the required country = group = file before or at the moment the first record of this group is being processed.