Date:         Sun, 13 Mar 2011 22:20:58 -0500
Reply-To:     Joe Matise <snoopy369@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Joe Matise <snoopy369@GMAIL.COM>
Subject:      Re: How can I replace the characters by a list in a fast way
Comments: To: Clark An <kuhasu@126.com>
In-Reply-To:  <773380d.9901.12eb25c8b64.Coremail.kuhasu@126.com>
Content-Type: text/plain; charset=UTF-8

Probably best to use one of the sort functions and then just iterate through
it deleting duplicates, I suppose.

-Joe

2011/3/13 Clark An <kuhasu@126.com>

>
> Thank you very much for your help.I will give it a try.
> The case is:
> an array with duplicate elements,some elements are not needed in a
> condiciton such as <50;
> but for the others, some of them are also not needed.
>
> And is there a way we can set up an array with not sure elements(I know we
> can do it in fcmp ,array[1] _temp_),and make sure there are no duplicate
> elements in the array?
>
> array orig[10] (1,1,2,2,2,3,5,1,1,2)
> array test[1] _temp_
> do i=1 to 10;
> j=1;
> test(j)=orig(i);/*How can I make sure test[*]can keep the non-duplicate
> elements*/
> j+1;
> end;
> end;
>
>
>
> At 2011-03-14 05:25:47，"Joe Matise" <snoopy369@gmail.com> wrote:
> I imagine a simple do loop would work of course ...
>
> do __i = 1 to dim(ff);
>  c = tranwrd(c,cats("ff(",__i,")"),ff[__i]);
> end;
>
> I don't think there is a CALL TRANWRD but if so that would be faster still.
>  Also you might test whether adding a IF FIND(...) would make it faster or
> not- some cost to doing the find/index but might save time over TRANWRD, if
> you only have a few instances in the string.
>
> If you want to improve over the do loop, particularly with a string like
> that (short), you could do it backwards.  Find the ( ) numbers, then replace
> them.  Something about like this probably would work.
>
> parencount = countc(c,')');
> do __i = 1 to parencount;
>  __arr = input(scan(c,2*__i,'()'),4.);
>  c = tranwrd(c,cats('ff(',__arr,')'),ff[__arr]);
> end;
>
> To improve even over that you could probably rebuild the string as you went
> instead of using the relatively expensive TRANWRD, but if you're only doing
> four or five of them per string you're probably okay.  However, if one ff(x)
> value can occur more than once in the string, you risk an error as the
> scan's iterator will stop tracking with the string - for production code you
> should put in checks for that.
>
> This might be faster, though I don't like having to separately strip off
> the 'ff'; I can't think of a good way to deal with it otherwise, though.
>
> parencount = countc(c,')');
> do __i = 1 to parencount;
>  __arr = input(scan(c,2*__i,'()'),4.);
>  before_arr = scan(c,(2*__i)-1,'()');
>  before_arr = substr(before_arr,1,length(before_arr)-2);
>  new_c = cat(new_c,before_arr, ff[__arr]);
> end;
>
> -Joe
>
>
> On Sun, Mar 13, 2011 at 10:49 AM, Clark An <kuhasu@126.com> wrote:
> array ff[250] (1,2,3....,250);
> c='Mail a ff(4) of the ff(1) to ff(3) and ff(14) with ff(8)';
>
> How can I use ff(x) to replace the 'ff(x)' in c string in a fast way pls?
> And if I can get a list 4 1 3 14 8,then in which way I need to store the
> list and then how to replace them(I mean if in a loop,but if there is a
> faster way without loop,better!)
>
> TIA!!
>
>
>
>
>
>