Sorting input records on length of characters.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Hi,

I have a file (LRECL=130,RECFM=FB) with names in it and want to sort it on first name's length. Is it possible with SYNCSORT?

Input:

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

Yes, what you have written should be possible, but your output seems to disagree that that is what you want.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

I don't understand you

Nic Clouston · Posted: Thu Nov 28, 2013 6:43 pm Post subject:

ramsri

Is your example input first name, initial, last name? The confusion arises because you have changed the sequence of the fields on the output but this is not mentioned in your requirement.

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

XAVIER is the second-shortest first name, but appears third in your expected output. Therefore, your expected output is not sorted on length of first name (the first "word" in your data).

dick scherrer · Posted: Fri Nov 29, 2013 8:28 am Post subject:

Hello,

What "rule" did you use to get from your input to your output?

Nic Clouston · Posted: Fri Nov 29, 2013 3:16 pm Post subject:

It looks as though your input is in surname/initial/forename and when you sort you transpose your data to be forename/initial./surname and sort on forename on output. But.. what if a person has no initial?

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

The rule is to sort names on first character on ascending order. Not seen a name without an initial in that dataset !

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

You want to sort on the first character of the third "word" on your input? And rearrange your data? Use PARSE and SQZ in INREC to rearrange your data. SORT FIELDS=(1,1,CH,A).

Obviously this will pickle with no initial, or with embedded blanks in a name (first or last) so ensure that you either can't have those or that you can deal with them.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Bill, your IDEA given me results -

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

Ramsri, you know that "it doesn't work" is useless without showing us what doesn't work.

dick scherrer · Posted: Mon Dec 02, 2013 6:07 am Post subject:

Hello,

You posted

hailashwin · New User Joined: 16 Oct 2008 Posts: 73 Location: Boston

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Ashwin, thanks.....I did not want to "do away" but wonder if it was possible...and you have shown that it was possible

Whenever I tried combination of INREC with OUTREC, I end up getting "INEFFICIENT USE OF INREC" warning in SYSOUT because I would have written but could not figure out.

Thanks again.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Dick, yes, I have deviated from what I originally asked....I wanted to SORT the names on their length after rearranging them.

Input-

Nic Clouston · Posted: Mon Dec 02, 2013 6:54 pm Post subject:

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

Ramsri,

Extend your records by a new field (INREC) that is the same as the maximum length of the names, populated by using your name field as the source for JFY with SHIFT=RIGHT. SORT on the new field. Looks like you want it descending. Chop the extension off afterwards (OUTREC or OUTFIL) with BUILD (of IFOUTLEN if you already happen to have IFTHEN processing).

Please try to have a complete, clear, question, with sample data which is representative and which exercises what you want to do.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Bill Woodger · DFSORT Moderator Joined: 09 Mar 2011 Posts: 7315

If you can't remember the DCB info for the datasets you are working on, then you need to use your fingers and eyes (type and read).

If you don't feel up to that you could always define your SORT key as each individual byte in your field, in reverse order, and remember that you seem to want, or seemed to want at the time of the particular post, descending.

This is getting nowhere, slowly. Unless something magical happens at your end, this topic will be locked.

ramsri · Active User Joined: 18 Oct 2008 Posts: 380 Location: India

Bill, I understand what has confused you.....Its not the length of dataset but the length of longest name in that dataset is the basis I am looking for.......ideas please.