Comparing two files and write unmatched fields to output fil

bhaskar_kanteti · Posted: Mon May 28, 2007 5:09 pm

Hi All,

I had a query in cobol.

I had two input files. Each input file will have 200 fields. These input files will use the same copy book. It is something like comparing the new file with yesterday's file.
Now i need to compare these two input files and which ever field doesn't match should only be written to output file.

Can any one help me how to do this logic.

i heard that there is some logic like comma seperator. I dont know exactly. If any one knows help me in this.

Regards,
Bhaskar.

krisprems · Posted: Mon May 28, 2007 5:20 pm

bhaskar_kanteti

give an example of i/p and o/p.

dick scherrer · Posted: Mon May 28, 2007 5:26 pm

Hello,

Comma separation (or any other delimiter) will not be part of your solution (unless the files are already delimited - which is doubtful).

In addition to comparing all of the fields to identify which have different values, you will also need to accomodate adds and deletes (which means there can be unmatched records in either file.

Imrankhan · New User Joined: 28 May 2007 Posts: 6 Location: Chennai

Hi Bhaskar,
plz follow these steps to compare 2 seq files:
1. sort 2 files with ur primary key. let primary keys be I1, I2
2. next read these 2 files;
read i1 (till end of file)
read i2 (till end of file)

3. process if I1 = i2
next sentence
else if I1 < I2
write unmatched records into output file
read I1(here key with lowest value should be read)
else if I1 > I2
write o/p file
read I2

this logic will definetly work cos i already done similar seq file comparison

bhaskar_kanteti · Posted: Tue May 29, 2007 8:49 am

The question is
I had two input files. Each input file will have 200 fields. These input files will use the same copy book. It is something like comparing the new file with yesterday's file.

Now i need to compare these two input files and when there is a mismatch in any of the field then write that particular filed from New File to output.

Suppose from the file below if NEW-FIELD-3 doesnot mattch with HIS-FIELD-3 then write only NEW-FIELD-3 to OF-FIELD-3

here is the FILE FORMAT.

Help me in this regard.

INPUT FILE 1:
NEW-FIELD-1 PIC X(10)
NEW-FIELD-2 PIC X(5)
NEW-FIELD-3 PIC X(4)
NEW-FIELD-4 PIC 9(3)
NEW-FIELD-5 PIC 9(8)
.
.
.
NEW-FIELD-200 PIC 9(12)

INPUT FILE 2:
HIS-FIELD-1 PIC X(10)
HIS-FIELD-2 PIC X(5)
HIS-FIELD-3 PIC X(4)
HIS-FIELD-4 PIC 9(3)
HIS-FIELD-5 PIC 9(8)
.
.
.
HIS-FIELD-200 PIC 9(12)

OUTPUT FILE:
OUT-FIELD-1 PIC X(10)
OUT-FIELD-2 PIC X(5)
OUT-FIELD-3 PIC X(4)
OUT-FIELD-4 PIC 9(3)
OUT-FIELD-5 PIC 9(8)
.
.
.
OUT-FIELD-200 PIC 9(12)

jaspal · New User Joined: 22 May 2007 Posts: 68 Location: mumbai

hi Bhaskar

why r u writing a cobol code go for ezytrieve,it will solve ur problem .
and it is easy

pingte · Active User Joined: 03 Dec 2005 Posts: 120 Location: india

hi,
to implement ur compare logic in cobol program u can follow this logic.
Step1:
in FD section declare copybook of old file as it is. then declare the copybook of new file (which is same as old file) with Replacing option (replacing option is used to differentiate between 2 files.. no impact on logic) also declare the record structure for o/p file
Step 2:
Open old and new file in i/p mode. open o/p file in Output mode
Read the old file till end of file.
for every record in old file read the new file (till EOF).
If record is found then move 'Y' to EOF flag for new file.
Else Write to o/p file.

Hope this helps.

bhaskar_kanteti · Posted: Tue May 29, 2007 6:47 pm

i think i should explain bit more clear.

First Input File:
-----------------
field-1 field-2 field-3 field-4 field-5 ..... field-200
210-----210------230-----320-----abc...........23.09 <--record1 of new i/p
110-----320------540-----230-----dlk............12.98 <--record2 of new i/p
and so on.....many number of records

Second Input File:
--------------------
field-1 field-2 field-3 field-4 field-5 ..... field-200
200-----110------230-----320-----xyz...........21.54 <--record1 of hist i/p
110-----520------540-----110-----xyz...........12.98 <--record2 of hist i/p
and so on.....many number of records

Output File:
-------------
field-1 field-2 field-3 field-4 field-5 ..... field-200
210------210----- -------- --------abc----------23.09
(this is the output by comparing record1 of new i/p with record1 of hist i/p)
----------320----- --------230-----dlk-----------
(this is the output by comparing record2 of new i/p with record2 of hist i/p)

hope this explanation is clear.
i cant go for any other option except cobol program.
kindly help me if any one know the logic for this.

dick scherrer · Posted: Tue May 29, 2007 9:16 pm

Hello,

If i correctly understand your requirement, the logic will not be complicated, but will be rather lengthy.

First, you need to set up the code to "match" both files by their "key" - your example does not show what the key might be. You have not said what should happen when there are entries in one file that are not in the other (i.e. adds and deletes).

Next your program needs to look at each field of the matched records (one by one) to see if the "new" field has the same value as the "old" field.

The output should be built one field at a time as the comparisons are made.

Bitneuker · Posted: Tue May 29, 2007 9:29 pm

I refuse to keep posting our wonderfull balance line solution. Done so several times

bhaskar_kanteti · Posted: Wed May 30, 2007 8:35 am

Hi,

In this both the new and hist files will have a key. If that key matches then only we will check for the rest of the fields and which ever is found mismatch we will write to output file from the new file. If the key doesn't match then ignore.

Can you tell me how to process this logic.

First Input File:
-----------------
key field-1 field-1 field-3 field-4 field-5 .......... field-200
aa --210-----210------230-----320-----abc...........23.09<-recd1 of new i/p
bb --110-----320------540-----230-----dlk............12.98<-recd2 of new i/p
and so on.....many number of records

Second Input File:
--------------------
key field-1 field-2 field-3 field-4 field-5 ........... field-200
aa --200-----110------230-----320-----xyz...........21.54<-recd1 of hist i/p
bb --110-----520------540-----110-----xyz...........12.98<-recd2 of hist i/p
and so on.....many number of records

Output File:
-------------
key field-1 field-2 field-3 field-4 field-5 ..... field-200
aa ---210------210----- -------- --------abc----------23.09
(this is the output by comparing record1 of new i/p with record1 of hist i/p)
bb --- ---------320----- --------230-----dlk-----------
(this is the output by comparing record2 of new i/p with record2 of hist i/p)

hope this explanation is clear.

pingte · Active User Joined: 03 Dec 2005 Posts: 120 Location: india

as i have said read file 1(old file) first .. take the record in cpybook layout..where u will know the fields defined...

when u read the second file(new file), for each record.. or based on key..
put record in same copybook layout (with replacing clause).. and compare it old record fields..

if non-matching fields are found put them into o/p record layout.

hope the xplanation is clear this time

Naresh Kumar Reddy P · New User Joined: 24 May 2007 Posts: 8 Location: Bangalore

Bhaskar,

1. Define a working storage table with 200 occurences and store the lengths of all your input 200 fields. Say it is defined as Field-Len(200)

2. Read File-new & File-old.

3. Perform step-4 until EOF reaches for either File-old or File-new

4. If key-new = key-old
initialize i, output-rec and move 1 to A
perform comare-para until i=200
write output-rec
else if key-new < key-old
write output-rec from new-rec
read File-new
else read File-old.

5. If File-new reaches EOF, end the process.
else read File-new and the write to output until File-new reaches EOF

Comare-para:
If (new-rec(A:Field-len(i)) = (old-rec(A:Field-len(i))
add Field-len(i) to A
add 1 to i
else
move new-rec(A:Field-len(i)) to output-rec(A:Field-len(i))
add 1 to i
End-if.

I'm sure this logic will definitely work. Let us know if you have any issues.

Thanks
Reddy

bhaskar_kanteti · Posted: Mon Jun 04, 2007 4:42 pm

THANK YOU SO MUCH NARESH KUMAR REDDY.........
THANKS FOR YOUR CLEAR EXPLANATION...........

THANK YOU SO MUCH...........