|
View previous topic :: View next topic
|
| Author |
Message |
ganeshprasanna
New User
Joined: 21 Feb 2009
Posts: 36
Location: South Portland, Maine
|
|
|
|
I have 2 variable length files with 1 ranging from 23 to 26543 and other ranging from 23 to 30623. Key for both files starts from 1 to 13 and of length 13 bytes.
I want to compare these 2 files and write an error report if the files are not matching. The mismatch can be of any type. ie; if any of the records are not matching, keys are not matching or if 1 file has more number of records than the other etc etc.
The file layouts are given below:
File 1:
| Code: |
01 WS-AAA-RECORD.
05 WS-AA-KEY PIC X(13).
05 WS-AA-VAR1 PIC X(08).
05 WS-AA-VAR2 PIC S9(04) COMP.
05 WS-AA-VAR1-INFO OCCURS 1 TO 2040 TIMES
DEPENDING ON
WS-AA-VAR2.
10 WS-AA-VAR3 PIC S9(04) COMP.
10 WS-AA-VAR4 PIC X(11). |
File 2:
| Code: |
01 WS-BBB-RECORD.
05 WS-BBB-KEY PIC X(13).
05 WS-BBB-VAR1 PIC X(08).
05 WS-BBB-VAR2 PIC S9(04) COMP.
05 WS-BBB-VAR1-INFO OCCURS 1 TO 2040 TIMES
DEPENDING ON WS-BBB-VAR2.
10 WS-BBB-VAR3 PIC S9(05) COMP.
10 WS-BBB-VAR4 PIC X(11). |
Please help me. |
|
| Back to top |
|
 |
Skolusu
Senior Member
Joined: 07 Dec 2007
Posts: 2205
Location: San Jose
|
|
|
|
ganeshprasanna,
1. Are these KSDS VSAM clusters?
2. Is it ok to just write out the keys in the report or do you want the entire records to be written out? |
|
| Back to top |
|
|
ganeshprasanna
New User
Joined: 21 Feb 2009
Posts: 36
Location: South Portland, Maine
|
|
|
|
Hi Skolusu,
1. Yes these are KSDS VSAM clusters.
2. Yeah. its ok with the key to be written in the file.
But can i also get exact posotion of mismatch also? It would be helpful. |
|
| Back to top |
|
|
Skolusu
Senior Member
Joined: 07 Dec 2007
Posts: 2205
Location: San Jose
|
|
|
|
ganeshprasanna,
The following DFSORT/ICETOOL JCL will give you the desired results
| Code: |
//STEP0100 EXEC PGM=ICETOOL
//TOOLMSG DD SYSOUT=*
//DFSMSG DD SYSOUT=*
//VSAM1 DD DSN=your vsam file 1,DISP=SHR
//VSAM2 DD DSN=your vsam file 2,DISP=SHR
//T1 DD DSN=&&T1,DISP=(MOD,PASS),SPACE=(CYL,(X,Y),RLSE)
//OUT DD SYSOUT=*
//TOOLIN DD *
COPY FROM(VSAM1) USING(CTL1)
COPY FROM(VSAM2) USING(CTL1)
SORT FROM(T1) USING(CTL2)
//CTL1CNTL DD *
INREC BUILD=(1,13)
OUTFIL FNAMES=T1,REMOVECC,HEADER1=(C'$')
//CTL2CNTL DD *
INREC IFTHEN=(WHEN=GROUP,BEGIN=(1,1,CH,EQ,C'$'),PUSH=(14:ID=1))
SORT FIELDS=(1,13,CH,A)
SUM FIELDS=(14,1,ZD)
OUTFIL FNAMES=OUT,OMIT=(1,1,CH,EQ,C'$'),
IFTHEN=(WHEN=(14,1,ZD,EQ,1),
OVERLAY=(14:C' - KEY FOUND ONLY IN FILE1')),
IFTHEN=(WHEN=(14,1,ZD,EQ,2),
OVERLAY=(14:C' - KEY FOUND ONLY IN FILE2')),
IFTHEN=(WHEN=(14,1,ZD,EQ,3),
OVERLAY=(14:C' - MATCHING KEY IN 2 FILES'))
/* |
This will create a file like this
| Code: |
KEY1 - KEY FOUND ONLY IN FILE1
KEY2 - KEY FOUND ONLY IN FILE2
KEY3 - MATCHING KEY IN 2 FILES
|
|
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hello,
Possibly, i misunderstand, but i believe the requirement includes comparing the entire record (and posting the position(s?) of the mismatch(es?) along with the key) when the keys match. . .
I wouldn't try to guess guess how duplicates should be handled. . .  |
|
| Back to top |
|
|
Skolusu
Senior Member
Joined: 07 Dec 2007
Posts: 2205
Location: San Jose
|
|
|
|
dick scherrer,
1. KSDS vsam clusters cannot have duplicates on KEY.
2. OP did say he is OK with just the keys. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hi Kolusu,
| Quote: |
| 1. KSDS vsam clusters cannot have duplicates on KEY. |
Oops, my bad - forgot we were talking vsam 
| Quote: |
| 2. OP did say he is OK with just the keys. |
That is the part i may have misunderstood. . . I believe he is ok with only the key being shown as output rather than entire records.
I believe he wants the key and any "data position" that does not match as the output - made more complicated as the elementary items in the arrays are not the same size. . .
Maybe he will clarify. |
|
| Back to top |
|
|
Skolusu
Senior Member
Joined: 07 Dec 2007
Posts: 2205
Location: San Jose
|
|
|
|
dick scherrer,
IF OP's Intention is to compare byte by byte then he needs to code a program to do it. DFSORT does not have the ability to compare byte by byte and pick the differences. |
|
| Back to top |
|
|
ganeshprasanna
New User
Joined: 21 Feb 2009
Posts: 36
Location: South Portland, Maine
|
|
|
|
Hi Kolusu,
Thanks for the quick response.
But as per my requirement, I need to compare the complete record and not just the keys.
But listing in the report can either be mismatched key or the complete record.
Is there any way out?
Thanks for all your inputs.
Thanks,
Ganesh |
|
| Back to top |
|
|
PeD
Active User
Joined: 26 Nov 2005
Posts: 459
Location: Belgium
|
|
|
|
Ecomp can do that easily.
Any chance you have it? |
|
| Back to top |
|
|
ganeshprasanna
New User
Joined: 21 Feb 2009
Posts: 36
Location: South Portland, Maine
|
|
|
|
Ped,
Could you please give me more Info on Ecomp?.
If its a third party tool, i am pretty sure we would not have it. |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6966
Location: porcelain throne
|
|
|
|
you know that in the two days you have been flumiging around looking for something easy (translate: written by someone else)
you could have coded a COBOL module to compare the 5 variables in each record.
use the match program from the "Programs" button on top,
using the match logic - which is the hardest to write
change what is necessary, write a perform for the 2040 occurances.
you can't use a generic compare
because the files are not the same
| Code: |
AAA record
10 WS-AA-VAR3 PIC S9(04) COMP.
BBB record
10 WS-BBB-VAR3 PIC S9(05) COMP.
|
|
|
| Back to top |
|
|
ganeshprasanna
New User
Joined: 21 Feb 2009
Posts: 36
Location: South Portland, Maine
|
|
|
|
Thanks Dick Brenholtz,
Of course COBOL module was my last resort. Wanted to know if there were any other options available. What about the runtime performance if COBOL module is used considering a large file with huge amount of records?
Any specific pointers to tune for performance? |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6966
Location: porcelain throne
|
|
|
|
The COBOL part, use INDEXED BY instead of subscripts.
The report should be a WRITE to a dataset instead of DISPLAY upon console.
Otherwise, the COBOL module should be about 100 lines (maybe more).
The VSAM tuning, I would suggest consulting the IBM REDBOOK VSAM Demystified
as well as search this forum.
Robert Sample (and others) have repeated posted on DCB parms that could/should be used. |
|
| Back to top |
|
|
PeD
Active User
Joined: 26 Nov 2005
Posts: 459
Location: Belgium
|
|
|
|
Ganesh,
| Quote: |
Ped,
Could you please give me more Info on Ecomp?.
If its a third party tool, i am pretty sure we would not have it. |
EComp is an independent software ( I invented ) which can cross compare any objects on IBM Mainframes : SAM, VSAM, DB2 tables, Queues, unloads od BD2 ( IBM, BMC ),extracts of SPUFI, QMF, SDSF, DDL's, Structures, apple, pear.... "Cross" means you can compare a SAM file with the content of a DB2 table, or a VSAM with a SAM, etc
The power of utility are in a nutshell : variety of criteria ( select, where clause, mutation, tolerance, reformat, timestamp neutralisation, ... and clear results : direct to the spot.
More documentation on request if really interested.
best regards
Pierre |
|
| Back to top |
|
|
dbzTHEdinosauer
Global Moderator
Joined: 20 Oct 2006
Posts: 6966
Location: porcelain throne
|
|
|
|
Pierre,
you site is unreachable without JAVA Script enabled,
which should be no needed. |
|
| Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006
Posts: 19244
Location: Inside the Matrix
|
|
|
|
Hi Pierre,
Does ecomp compare elements with the same logical content, but differing data lengths?
For example 000000123 and 00123 should compare equal even though they are different lengths. . . And these are embedded within arrays in the 2 files. |
|
| Back to top |
|
|
PeD
Active User
Joined: 26 Nov 2005
Posts: 459
Location: Belgium
|
|
|
|
| Quote: |
Does ecomp compare elements with the same logical content, but differing data lengths?
For example 000000123 and 00123 should compare equal even though they are different lengths. . . And these are embedded within arrays in the 2 files.
|
Yes Ecomp can do that on demand : 00000123 can be equal to 123,
as 123 in packed decimal can be equal to 123 in numeric.
It is an option.
Ecomp is also able to compare records with structure variables as "depending on" in Cobol or "refer block" in PL/I.
example : count "0003 abc def ghi" compared to "0002 abc ghi" will show only difference on def and accept ghi even if the positions in the records is not same depending on the counter.
And more and more...
To Dick ( somewhere in Germany )
| Quote: |
Pierre,
you site is unreachable without JAVA Script enabled,
which should be no needed.
|
Yes I have to work on that urgently.
Thanks
Pierre |
|
| Back to top |
|
|
|
|
