|
View previous topic :: View next topic
|
| Author |
Message |
tamminenisidhartha
Currently Banned
New User
Joined: 31 Oct 2012
Posts: 43
Location: INDIA
|
|
|
|
I have the input file like this.
| Code: |
emp dep emp type emp sal
000011 P 100
000011 T 200
000011 S 500
000023 P 300
000024 P 400 |
I would like to know this. I want to eliminate only emptype " P and T " records from writing if they have same emp dep value.
| Code: |
Note : In this example, The Emptype for first two records is " P and T ".
And they have the same empdep value.
Here the EMPTYPE " S" also has the same empdep value.
But it should be written into the output file.
So I would like to avoid only this pair of record from writing into
the output file. |
Here's he logic on what I have been working on.
| Code: |
IF K = SUBSTR(INREC.EMPNO,1,6) THEN DO;
IF P ¬= SUBSTR(INREC.TYPE1,1,4) THEN DO;
WRITE FILE(OUTFILE) FROM (INREC);
END;
ELSE DO;
WRITE FILE(OUTFILE) FROM (HEADER);
WRITE FILE(OUTFILE) FROM (INREC);
END;
END;
END;
K = SUBSTR(INREC.EMPNO,1,6);
P = SUBSTR(INREC.TYPE1,1,4);
READ FILE(INFILE) INTO (INREC); |
I have coded this inside a do while loop.
I hope you understood my question.
Code'd |
|
| Back to top |
|
 |
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011
Posts: 7309
Location: Inside the Matrix
|
|
|
|
Your post has been Code'd to preserve formatting. Please do this yourself in future.
You will need to store the P/T records as you only know by the receipt of a T after the P (or P after the T) that the P (or T) is not needed for that key.
What's all the SUBSTR for? |
|
| Back to top |
|
|
tamminenisidhartha
Currently Banned
New User
Joined: 31 Oct 2012
Posts: 43
Location: INDIA
|
|
|
|
| I have been comparing the values by passing them into the variables using substring function. |
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10903
Location: italy
|
|
|
|
but why not use a proper DCL for the record layout
if the layout changes, You will not to have to chase around for the offsets |
|
| Back to top |
|
|
Akatsukami
Global Moderator
Joined: 03 Oct 2009
Posts: 1787
Location: Bloomington, IL
|
|
|
|
| enrico-sorichetti wrote: |
but why not use a proper DCL for the record layout
if the layout changes, You will not to have to chase around for the offsets |
The data the TS shows suggests that the records are not necessarily of a fixed format (although heesh might have munged the example by putting explanatory column headers in the Code tags).
Tammi-kun, are more employee types possible than "P", "S", and "T"? May the data be assumed to be sorted by department? |
|
| Back to top |
|
|
tamminenisidhartha
Currently Banned
New User
Joined: 31 Oct 2012
Posts: 43
Location: INDIA
|
|
|
|
| The data is srted by EMPDEP already and the employee types are only of P, S, T types. |
|
| Back to top |
|
|
Akatsukami
Global Moderator
Joined: 03 Oct 2009
Posts: 1787
Location: Bloomington, IL
|
|
|
|
Very well. Consider the following program:
| Code: |
foo36: proc options (main) reorder;
%dcl (true,false,digits,space) char;
%true = '''1''B';
%false = '''0''B';
%digits = '''0123456789''';
%space = ''' ''';
dcl tulin file record input
env (fb recsize(80)),
tulout file record output
env (fb recsize(132));
dcl eyecatch (11) unsigned fixed bin (32) static
init ('E69989A3'XU, 'A3859540'XU, '82A840C1'XU,
'9281A3A2'XU, 'A4928194'XU, '8960A281'XU,
'948140C8'XU, '8589A285'XU, '8940F2F4'XU,
'4BF1F14B'XU, 'F1F50000'XU);
dcl (eof_tulin, set_aside) bit init(false),
emptyp char,
(empdep, lastdep) char (79),
(recin, recaside) char (80) init((80)' '),
recout char (132),
(p, q) fixed bin (31);
dcl (index, substr, verify) builtin;
on endfile (tulin) eof_tulin = true;
read file (tulin) into (recin);
do while (¬eof_tulin);
if (verify(substr(recin,1,1),digits)¬=0) then
do;
recout = recin;
write file (tulout) from (recout);
end;
else do;
p = index(recin,space);
q = verify(recin,space,p);
empdep = substr(recin,1,p-1);
emptyp = substr(recin,q,1);
if (empdep¬=lastdep) then
do;
if (set_aside) then
do;
recout = recaside;
write file (tulout) from (recout);
end;
set_aside = false;
lastdep = empdep;
end;
select (emptyp);
when ('S') do;
recout = recin;
write file (tulout) from (recout);
end;
when ('P', 'T') do;
recaside = recin;
set_aside = ªset_aside;
end;
otherwise do;
put skip edit ('Serious error! Unknown EMPTYP = [', emptyp,
']') (a,a,a);
exit;
end;
end;
end;
read file (tulin) into (recin);
end;
if (set_aside) then
do;
recout = recaside;
write file (tulout) from (recout);
end;
end foo36;
|
I have assumed that the first character of EMPDEP is always a decimal digit; that criterion may be used to eliminate header records from consideration. I further assume that, although EMPDEP is always separated from EMPTYP by at least one space (and that it contains no embedded spaces), the record is not of fixed format; thus the use of the INDEX and VERIFY functions to find EMPTYP (if the record is of fixed format, you are better off declaring a structure as suggested by Dr. Sorichetti). An EMPTYP of "P" or "T" causes the record to be set aside and the switch SET_ASIDE to be toggled; when EMPDEP changes, the set-aside record is written if SET_ASIDE is true, which it will not be if a pair of records is encountered. Note that signficantly different data may require changes. |
|
| Back to top |
|
|
tamminenisidhartha
Currently Banned
New User
Joined: 31 Oct 2012
Posts: 43
Location: INDIA
|
|
|
|
@akatsukami.
Thank you.
I would execute the code and post the outcome here. |
|
| Back to top |
|
|
|
|
  |
All times are GMT + 6 Hours |
|
