|
View previous topic :: View next topic
|
| Author |
Message |
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
Hi.
Is there anyway to handel to error caused by a function call ?
I am using the NUMVAL-C function, but it happened that the string contains some digit after the first space. So it cause a U4038 abend :
Unlimited : < N > (Y/N)
Amount : < 1 15,00 > <-- encoded value
This is the message found in the IMS region :
A non-digit character other than a decimal point, comma, space or sign
(+,-,CR,DB) was found in argument-1 for the NUMVAL/NUMVAL-C function.
The IBM message that corresponds to the condition is:
IGZ0152S Invalid character 5 was found in column 17 in argument-1 for
function NUMVAL-C in program GUF71 at displacement X'5D46'.
Thanks for your answer. |
|
| Back to top |
|
 |
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
| eric henken wrote: |
Hi.
Is there anyway to handel to error caused by a function call ?
|
I think there is no Error Handling for function call in COBOL. There are very few Error Handling available in COBOL, like ON SIZE ERROR.
We can convert only Numeric Values(Including Spaces, decimal Point, currency sign) present in alphanumeric variables to Numeric using NUMVAL-C function.
Look at NUMVAL-C Details |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6966
Location: porcelain throne
|
|
|
|
NUMVAL (and -C) obviously have limitations. according to the documentation, there are some restrictions.
you can not use it on a garbage field. it was designed to take x-type input and convert to numeric. if you have digit-space-digit-digit you have two fields
so, a little cleanup is necessary.
declaratives will stop the error going to mvs.
you clean up the field or use reference modfication to define you source for the NUMVAL Function. |
|
| Back to top |
|
|
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
Thanks for your answer.
I was trying to do this without to have to clean up the field 
I have already define a SPECIAL NAME CLASS :
| Code: |
SPECIAL-NAMES.
CLASS TEST-NUM-E '0' THRU '9' ',' '.' ' ' |
to be able to validate the contents of the field before to use the function, but in a class we can't say where the spaces must be placed.
I am going to eliminate the spaces and move the result to another variable.
Regards,
Eric. |
|
| Back to top |
|
|
Bill O'Boyle
CICS Moderator
Joined: 14 Jan 2008
Posts: 2501
Location: Atlanta, Georgia, USA
|
|
|
|
A brilliant move on IBM's part would have been to create an "LE" (Language Environment) version of NUMVAL and NUMVAL-C, which would then allow the Caller to check the feedback-code halfword for validity and successful completion.
Unfortunately, unlike (as an example) many of the COBOL Date FUNCTIONS which have corresponding "LE" Callable Service routines, they didn't do that and the Caller is forced to ensure data-validation before using either of these FUNCTIONS and others as well.
Regards,
Bill |
|
| Back to top |
|
|
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
| eric henken wrote: |
I am going to eliminate the spaces and move the result to another variable. |
You can use INSPECT,Reference Modification to eliminate SPACES in your alphanumeric variable. |
|
| Back to top |
|
|
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
Thank all !
I solved my problem with the following solution.
| Code: |
MOVE 0 TO J-IND
MOVE SPACES TO WS-AMOUNT-X
PERFORM VARYING I-IND FROM 1 BY 1
UNTIL I-IND > 20
IF GUF71-AMOUNT (I-IND:1) > SPACES
ADD 1 TO J-IND
MOVE GUF71-AMOUNT (I-IND:1) TO WS-AMOUNT-X (J-IND:1)
END-IF
END-PERFORM
MOVE WS-AMOUNT-X TO GUF71-AMOUNT
|
So I can shift all digits to the left and call the function without any spaces between digit !
I hope that IBM will develop tools to help us to be more efficient in the handling of error !
Regards,
Eric. |
|
| Back to top |
|
|
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
| Thanks Eric for writing your solution here. |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6966
Location: porcelain throne
|
|
|
|
eric,
I wonder why there was a space between the 1 and the 15,00?
IBM provides in its instruction-set commands necessary to process business data.
the commands available, had they been properly used during creation of this x-type field, would have made it a very easy NUMVAL to extract the data. But, the field was populated in an unprofessional manner. spilting up a number with spaces intersperced is really bad. are you sure that it is 115,00 and not 15,00. I assume your environment is COMMA IS DECIMAL. you could be heading for a 100 point error.
Don't hold your breath waiting for IBM to build an instruction to take care of poorly generated data. |
|
| Back to top |
|
|
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
Hi Dick,
The data is coming from an input field of a screen.
I develop application with a tool and it doesn’t work well with numeric field!
The value must be displayed in a edited format PIC Z.ZZZ.ZZZ.ZZZ.ZZ9,99 thus allow spaces, digits, points and comma. The user can introduce what he wants and I must validate that what he give is correct.
I can't modify the MFS generated by the tools, so I can't say to IMS that the input field is numeric and the output field is char.
So I have searched other solution to be able to manipulate this field without writing a lot of Cobol Instruction.
Eric.
I have found the NUMVAL and NUMVAL-C function and hope that it was possible to handle the error before the program dump ! I was dreaming....
This is the reason why the input field can be fill in the improper way. |
|
| Back to top |
|
|
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
Eric, in your case can user enters values like 1150BB where B is for BLANK?
If user can enter like this, then you will be in trouble & again have to take help of INSPECT & Reference Modification.
I had worked with similar kind of task before, where I had to take care of all probable cases. |
|
| Back to top |
|
|
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
Yes, the user can introduce 1150BB !
But, if you take a look at the NUMVAL-C function, you see that the input string can contain spaces.
To test this, I have make the following test
| Code: |
SPECIAL-NAMES.
CLASS TEST-ERIC '0' THRU '9' '.' ',' ' '
DATA DIVISION.
FILE SECTION.
FD PARMCARD
LABEL RECORDS ARE OMITTED
RECORDING MODE IS F
RECORD CONTAINS 80
BLOCK CONTAINS 0
DATA RECORD IS REC-PARMCARD.
01 REC-PARMCARD.
05 FUNC-REC PIC X(10).
05 DATAS PIC X(70).
WORKING SECTION.
01 I-IND PIC S9(04) COMP.
01 J-IND PIC S9(04) COMP.
01 I-CTR PIC S9(04) COMP.
01 WS-TESTX PIC X(10).
01 WS-TEST2 PIC 9(10)V9(02).
01 WS-TEST2-EDT PIC ZZ.ZZZ.ZZ9,99.
PROCEDURE DIVISION
MOVE 0 TO I-CTR
INSPECT DATAS(1:10)
TALLYING I-CTR FOR ALL ','
IF I-CTR > 1
DISPLAY 'GUTSTEH' ' - ' 'DATA NOT TEST-ERIC COMPLIANT'
DATAS(1:10)
ELSE
MOVE 0 TO J-IND
MOVE SPACES TO WS-TESTX
PERFORM VARYING I-IND FROM 1 BY 1
UNTIL I-IND > 10
IF DATAS(I-IND:1) > SPACES AND DATAS(I-IND:1) NOT = '.'
COMPUTE J-IND = J-IND + 1
MOVE DATAS(I-IND:1) TO WS-TESTX (J-IND:1)
END-IF
/* Ending: IF DATAS(I-IND:1) > SPACES AND DATAS(I-IND:1) NOT =...
END-PERFORM
MOVE WS-TESTX TO DATAS(1:10)
IF DATAS(1:10) IS TEST-ERIC AND DATAS(1:10) > SPACES
COMPUTE WS-TEST2 = FUNCTION NUMVAL-C(DATAS(1:10))
MOVE WS-TEST2 TO WS-TEST2-EDT
DISPLAY 'GUTSTEH' ' - ' 'WS-TEST2*'
WS-TEST2
'*'
DATAS(1:10)
'*WS-TEST2-EDT*'
WS-TEST2-EDT
'*'
ELSE
DISPLAY 'GUTSTEH' ' - ' 'DATA NOT TEST-ERIC COMPLIANT *
''
DATAS(1:10)
'*'
END-IF
/* Ending: IF DATAS(1:10) IS TEST-ERIC AND DATAS(1:10) > ...
END-IF.
/* Ending: IF I-CTR > 1...
|
The input value are :
| Code: |
3.000,00
3 3.000,00
3.000,00
3,00
3
3,,,
....3
A3
B --> Blank
|
It gives as result :
| Code: |
GUTSTEH - WS-TEST2*000000300000*3000,0 *WS-TEST2-EDT* 3.000,00*
GUTSTEH - WS-TEST2*000003300000*33000,0 *WS-TEST2-EDT* 33.000,00*
GUTSTEH - WS-TEST2*000000300000*3000,00 *WS-TEST2-EDT* 3.000,00*
GUTSTEH - WS-TEST2*000000000300*3,00 *WS-TEST2-EDT* 3,00*
GUTSTEH - WS-TEST2*000000000300*3 *WS-TEST2-EDT* 3,00*
GUTSTEH - DATA NOT TEST-ERIC COMPLIANT3,,,
GUTSTEH - WS-TEST2*000000000300*3 *WS-TEST2-EDT* 3,00*
GUTSTEH - DATA NOT TEST-ERIC COMPLIANT *A3 *
GUTSTEH - DATA NOT TEST-ERIC COMPLIANT * * |
So you see, with the code that I have added to shift the input field to the left, I think that I can expect everything from the user 
At first the field must be validate :
1) One comma allowed
2) Remove blank and point (It give an error if there is leading point !)
3) validate that the field is TEST-ERIC compliant (the point may be removed in the CLASS)
Then use the function to format the input field !
I think from now it will be OK.
Eric |
|
| Back to top |
|
|
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
Yeah, I also hope this will work.
In my task there was one more case of 'LEADING ZEROS' (like LEADING SPACES) say if user enters
We can overcome to this one also.
Regards,
Yogeshwar |
|
| Back to top |
|
|
eric henken
New User
Joined: 04 May 2007
Posts: 6
Location: Brussel
|
|
|
|
It works also with severals leading zeroes
| Code: |
GUTSTEH - WS-TEST2*000000000300*0000003 *WS-TEST2-EDT* 3,00*
|
If I have forgotten something, I will give the modification ! |
|
| Back to top |
|
|
yogeshwar_ade
Active User
Joined: 31 Aug 2006
Posts: 103
Location: INDIA
|
|
|
|
Eric,
Thanks for all your explanation.
You covered all possible cases. I dont think USER will have any other possible case which will cause your code ABEND.
Regards,
Yogeshwar |
|
| Back to top |
|
|
|
|
