IBM Mainframe Forum Index
 
Log In
 
IBM Mainframe Forum Index Mainframe: Search IBM Mainframe Forum: FAQ Register
 

Handling the numeric data in unstring in COBOL


IBM Mainframe Forums -> COBOL Programming
Post new topic   Reply to topic
View previous topic :: View next topic  
Author Message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Sun Jan 02, 2022 8:38 pm
Reply with quote

Hi All,

I was trying to read the CSV file format in Cobol , by unstringing the data which comes with a comma delimited .

In the input data i get the numeric data like 99.90. So after i do unstring to write to a output file i want in the format 999{ which is hex for ebcdic.

Hoe we can convert the data from decimal to a non decimal format

Please let me know
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Sun Jan 02, 2022 9:19 pm
Reply with quote

It would be nice to know: what is “non decimal format”?
As well as: which one is “decimal”?

PIC 99999
PIC 99999 COMP-3
PIC 99999 COMP
PIC 99999 COMP-n
PIC XXXXX
. . . etc . . .
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Sun Jan 02, 2022 9:52 pm
Reply with quote

FYI:

COBOL Numeric Formats
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Mon Jan 03, 2022 7:09 am
Reply with quote

@Sergeyken

The Data comes from a SQL Server download thru a CSV File to mainframe.

So in the Data base it will be like Dec(7,2)

and in mainframe copybook we unstring this to S9(07)V99
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Mon Jan 03, 2022 4:44 pm
Reply with quote

We were actually trying to make it like if the input comes

input file >> 18.50

output file should write it like 185}
which means it should be hex at the end
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Mon Jan 03, 2022 7:05 pm
Reply with quote

vibi wrote:
@Sergeyken

The Data comes from a SQL Server download thru a CSV File to mainframe.

So in the Data base it will be like Dec(7,2)

and in mainframe copybook we unstring this to S9(07)V99

All fields in a CSV file are strictly in character format. That is, PIC X(nnn) in COBOL.
After your DB field DEC(7,2) has been unloaded from a database to CSV format, is is expected to be in PIC X(9) format; please, verify your real unloaded data to be sure, before trying any conversion.
For instance, the following values are expected to be unloaded as follows
Code:
-123.45 ===> "  -123.45" (with two spaces in front of the field)
+123.45 ===> "   123.45" (with three spaces in front of the field)

Using the link given above, please find out: how to convert in COBOL
PIC X(nnn) ===> PIC S9(07)V99
This must be one of your basic knowledge before starting any coding.

All such very fundamental knowledge is expected to come from technical manuals, rather than from forums.
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Wed Jan 05, 2022 8:03 am
Reply with quote

Thank you Sergeyken

I want the decimal value on the CSV data ( for example 12.20) stored as 122{ in the output file...right now it is stored as 12.20

Usually it will be like that , but not sure why it was not writing as 122{ after the unstring
Back to top
View user's profile Send private message
Garry Carroll

Senior Member


Joined: 08 May 2006
Posts: 1140
Location: Dublin, Ireland

PostPosted: Wed Jan 05, 2022 1:38 pm
Reply with quote

Does your source data have a positive/negative sign?

1220 is neither positive or negative and is generally treated as positive.

122{ is positive while 122} is negative.

Garry.
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Wed Jan 05, 2022 1:47 pm
Reply with quote

hi Gary,

data from the csv will come like decimal 10.20...
but when we unstring it it is coming as same thought we unstring this to 9(02)v99

I want the output file to like 102{ to match an existing format
Back to top
View user's profile Send private message
Garry Carroll

Senior Member


Joined: 08 May 2006
Posts: 1140
Location: Dublin, Ireland

PostPosted: Wed Jan 05, 2022 1:59 pm
Reply with quote

Have you tried unstring to a field with the (optional) sign specified ( e.g. PIC S9(02)V99 ) ?

Garry
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Wed Jan 05, 2022 2:01 pm
Reply with quote

Yes Gary , I am unstringing to a variable S9(02) V99

I was expecting to come in the format i expected.. like 102{

My understanding is when we unstring it to S99v99 then the decimal portion wont come in the output file...

But right now it is coming like 10.20 itself
Back to top
View user's profile Send private message
Garry Carroll

Senior Member


Joined: 08 May 2006
Posts: 1140
Location: Dublin, Ireland

PostPosted: Wed Jan 05, 2022 2:18 pm
Reply with quote

I'm GaRRy.

It would seem that the values from the .csv file are all unsigned and you want to consider as signed positive. You might try adding +0 to the field to get the sign set to what you want.

Garry.
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Wed Jan 05, 2022 6:41 pm
Reply with quote

Garry Carroll wrote:
Does your source data have a positive/negative sign?

1220 is neither positive or negative and is generally treated as positive.

122{ is positive while 122} is negative.

Garry.


Any field unloaded as CSV has mandatory the sign-character as part of it (while for positive values it may be either blank, or '+'). The position of the sign within this field is varying. It is almost free-text format.

There is no such problem when DB records unloaded in flat-file format.

I asked the TS about his actual format, and what are his intentions on handling varying positions of the sign, and leading blanks? My questions were simply ignored.

Conversion of free-text numbers to decimal unpacked numerics requires either a conversion function (if available), or coding a mini text analyzer inside of COBOL code.
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Wed Jan 05, 2022 6:51 pm
Reply with quote

vibi wrote:
Thank you Sergeyken

I want the decimal value on the CSV data ( for example 12.20) stored as 122{ in the output file...right now it is stored as 12.20

Usually it will be like that , but not sure why it was not writing as 122{ after the unstring

I know what you mean, but please keep in mind that CSV-formatted numerics (e.g. DEC(7,2) ) do have non-fixed positions of the sign character, as well as leading blanks in front of the numeric value. This is your original problem: free-text parsing. How do you plan to detect negative values in your CSV? You need also to ignore decimal point character somehow, etc.
Code:
-12345.67
 12345.67
   123.45
  -123.45
     0.12
    -0.12
. . . etc ...
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Wed Jan 05, 2022 7:51 pm
Reply with quote

Thank You ,

What i was able to do is unstring the decimal numerics into alphanumeric and then later move the alpha to the S9(07)V99 and then pass it thru a NUMVAL Function where i am able to get the unpacked decimal properly ..

Regarding the negative i am still looking that options.

We are actually creating a file which needs to be passed to downstream application. Earlier there was stored procedure in SQL server which was sending the data to mainframe but now as the DB is moved to cloud they are going to send the data in the CSV format. SO the main intention is to create a flat file similar to the Stored procedure created by SQL server.
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Wed Jan 05, 2022 8:08 pm
Reply with quote

Off-topic.

Has this COBOL program any other purpose except this conversion of CSV format(s) to regular tabulated numeric formats?

If not, the best solution would be - using any of SORT utilities for any desired conversion, including full free-text analysis.
Back to top
View user's profile Send private message
vibi

New User


Joined: 02 Jan 2022
Posts: 8
Location: India

PostPosted: Wed Jan 05, 2022 8:33 pm
Reply with quote

Yes.. first i tried JCl sort to parse but there more than 1000+ in table.. which i think SORT has limitation ..

COBOl module is just to parse and write to a output file
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Wed Jan 05, 2022 10:19 pm
Reply with quote

vibi wrote:
Yes.. first i tried JCl sort to parse but there more than 1000+ in table.. which i think SORT has limitation ..

COBOl module is just to parse and write to a output file


More than 1000+ of what??? Characters? Fields? Records? Datasets? Volumes?

SORT utilities do have much less limitations, and much better performance compared to almost any COBOL code.
In many cases the size of code for SORT utility is 50-100 times less than equivalent COBOL code.
Back to top
View user's profile Send private message
sergeyken
Warnings : 2

Senior Member


Joined: 29 Apr 2008
Posts: 1413

PostPosted: Wed Jan 05, 2022 11:39 pm
Reply with quote

Code:
//CONVERT EXEC PGM=SORT
//SYSOUT   DD  SYSOUT=*
//SORTIN   DD  *    - unloaded DEC(7,2) as CSV
. . . . , "    -123.45", . . . .
. . . . . . . . . . , " +98765.12", . . . .
. . , " 67123.45", . . . .
//*
//SORTOUT  DD  SYSOUT=*
//SYSIN    DD  *
 INREC PARSE=(%=(STARTAFT=C','),
              %1=(STARTAFT=C'"',ENDBEFR=C'"',
                  FIXLEN=10)),
       BUILD=(%1,SFF,TO=ZDC,LENGTH=7)  as if PIC S9(5)V99
 SORT FIELDS=COPY
 END
//*
Back to top
View user's profile Send private message
View previous topic :: :: View next topic  
Post new topic   Reply to topic View Bookmarks
All times are GMT + 6 Hours
Forum Index -> COBOL Programming

 


Similar Topics
Topic Forum Replies
No new posts File Handling COBOL Programming 9
No new posts Decompress data from JMR log file Compuware & Other Tools 4
No new posts Search substring in internal table of... COBOL Programming 2
No new posts Transfer TB of data from Mainframe to... All Other Mainframe Topics 2
No new posts Calling COBOL DB2 program from a COBO... COBOL Programming 2
Search our Forums:

Back to Top