REXX/SORT split sequential file

Question

I need to split a sequential Mainframe file. Well, to be precise I need to copy the content from this file to another, starting from a specific keyword. Example:

line1
line2
line3
start line4
line5
line6

In this case I need to search for "start" and copy everything starting from line4 to another file using either REXX or SORT. Any suggestions?

EDIT: What I thought about but not happy with in REXX

"EXECIO * DISKR INPUT (STEM INPUT. FINIS)" 
LINEINPUT = 1 
LINEOUTPUT = 1 
FOUND = 0 /*working like a boolean? 
DO WHILE LINEINPUT <= INPUT.0 
  IF INPUT.LINEINPUT = start line4 THEN DO 
  FOUND = 1 
  END 
  IF FOUND = 1 THEN DO 
    INPUT.LINEINPUT = OUTPUT.LINEOUTPUT 
    LINEOUTPUT = LINEOUTPUT + 1 
  END 
LINEINPUT = LINEINPUT + 1 
END

Something like this maybe, but this means that I need to go through all those files line by line. Maybe there is a better way in JCL? Maybe Syncsort can do something like this?

RECFM is fixed. FBA to be precise. LRECL 170. The trigger can either be a part of this line or the whole line. This is not important, because it is always the same line in every single file and even the first few characters are unique in the whole file. Means that "start" only appears once in the file.

OK. You can delete all your comments now. Hover over a comment you have added, and you will see an x appear at the end of the line. Click on that. All the information is now in your question, so the comments, which are ephemeral here, can go. — Bill Woodger, Jan 05 '15 at 14:16

score 4 · Accepted Answer · answered Jan 05 '15 at 10:34

For a fixed-position start of the trigger:

  OPTION COPY
  INREC IFTHEN=(WHEN=GROUP,
                 BEGIN=(1,5,CH,EQ,C'start'),
                 PUSH=(171:ID=1))
  OUTFIL OMIT=(171,1,CH,EQ,C' '),
         BUILD=(1,170)

For a variable-position, unique, trigger:

  OPTION COPY
  INREC IFTHEN=(WHEN=GROUP,
                 BEGIN=(1,170,SS,EQ,C'start'),
                 PUSH=(171:ID=1))
  OUTFIL OMIT=(171,1,CH,EQ,C' '),
         BUILD=(1,170)

WHEN=GROUP gives you PUSH, which will put data from the current record, or a group-number (ID) or number within group (SEQ), to that position on all the records in the group (including the current record). In this case, the "group" is the rest of the file.

The SS is field-type allowing a Sub-String search.

Then an OMIT= (similar to OMIT COND= but after the file has been processed by the control cards) on OUTFIL does the actual selection from the flag (the automatically extended records not with a value from the PUSH will be set to blank) and BUILD to drop off the extra byte.

The extra byte is needed because SORT has no "program storage" for definitions - extra fields have to be on the records (or, for the duration of the current record only, in PARSEd fields).

score 0 · Answer 2 · answered Jan 05 '15 at 09:07

Sorry for not commenting as i dont have enough reputation i am writing this in answer. Will remove this later

You can achieve this by both the methods i.e. Through REXX or JCL. Please show us what you tried till now so we can figure out what to improve.

REXX/SORT split sequential file

2 Answers2