Date:         Mon, 3 Mar 2008 11:49:59 -0500
Reply-To:     "Howard Schreier <hs AT dc-sug DOT org>"
              <schreier.junk.mail@GMAIL.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "Howard Schreier <hs AT dc-sug DOT org>"
              <schreier.junk.mail@GMAIL.COM>
Subject:      Re: AW: read out a string

On Mon, 3 Mar 2008 09:16:18 -0500, Peter Crawford
<peter.crawford@BLUEYONDER.CO.UK> wrote:

>On Mon, 3 Mar 2008 09:02:59 -0500, Howard Schreier <hs AT dc-sug DOT org>
><schreier.junk.mail@GMAIL.COM> wrote:
>
>>On Mon, 3 Mar 2008 12:40:18 +0100, Stefan Pohl <stefan.pohl@ISH.DE> wrote:
>>
>>>Dear Howard,
>>>
>>>how is your code to modify that it can handle this input
>>>
>>>data have1;
>>>   input ids $50.;
>>>   cards;
>>>[1 226] [227 390] [391 419] [420 435] [436 439]
>>>[1 226] [227 390] [391 435] [436 439]
>>>[1 226] [227 390] [391 439]
>>>[1 390] [391 439]
>>>[1 439]
>>>   ;
>>>run;
>>>
>>>Thank you for your help.
>>
>>Short answer: Change the F informat to D.
>>
>>Explanation: The documentation for the w.d informat states that BESTw.d,
>>Dw.d, Ew.d, Fw.d are all aliases for w.d. However, it appears to F has a
>>default width of 1 whereas the other 3 have default width 12. I run this
>code:
>>
>>   data _null_;
>>   best = input('1234567890123',best.);
>>      d = input('1234567890123',   d.);
>>      e = input('1234567890123',   e.);
>>      f = input('1234567890123',   f.);
>>   put +3 _all_;
>>   run;
>>
>>I see
>>
>>   best=123456789012 d=123456789012 e=123456789012 f=1 _ERROR_=0 _N_=1
>>
>>So the F informat is not exactly an alias for the others.
>>
>>>
>>>Stefan.
>>>
>>>
>>>
>>>> I agree with Gerhard. Put aside macro for now, and
>>>> concentrate on the DATA step and/or PROC code needed to solve
>>>> the problem. Then go back and macro-ize if there is really
>>>> need for that.
>>>>
>>>> The notation with square brackets and concatenation of
>>>> multiple pairs may be familiar from the literature, but it is
>>>> not useful in SAS. So first extract all of the values into a
>>>> nice data structure. Here's one way:
>>>>
>>>>    data have;
>>>>    input ids $50.;
>>>>    cards;
>>>> [1 2] [3 3] [4 4] [5 5] [6 6] [7 7] [8 8] [9 9]
>>>> [1 4] [5 5] [6 6] [7 7] [8 8] [9 9]
>>>> [1 4] [5 5] [6 8] [9 9]
>>>> [1 9]
>>>>    ;
>>>>
>>>>    data need(keep = i j id1 id2);
>>>>    set have;
>>>>    i = _n_;
>>>>    ids = compress(tranwrd(ids,'] [','|'),'[]');
>>>>    do j = 1 by 1;
>>>>       pair = scan(ids,j,'|');
>>>>       if missing (pair) then leave;
>>>>       ID1 = input(scan(pair,1,' '),f.);
>>>>       ID2 = input(scan(pair,2,' '),f.);
>>>>       output;
>>>>       end;
>>>>    run;
>>>>
>>>> Result:
>>>>
>>>>    Obs    i    j    ID1    ID2
>>>>
>>>>      1    1    1     1      2
>>>>      2    1    2     3      3
>>>>      3    1    3     4      4
>>>>      4    1    4     5      5
>>>>      5    1    5     6      6
>>>>      6    1    6     7      7
>>>>      7    1    7     8      8
>>>>      8    1    8     9      9
>>>>      9    2    1     1      4
>>>>     10    2    2     5      5
>>>>     11    2    3     6      6
>>>>     12    2    4     7      7
>>>>     13    2    5     8      8
>>>>     14    2    6     9      9
>>>>     15    3    1     1      4
>>>>     16    3    2     5      5
>>>>     17    3    3     6      8
>>>>     18    3    4     9      9
>>>>     19    4    1     1      9
>>>>
>>>> On Fri, 29 Feb 2008 04:02:47 -0500, Gerhard Hellriegel
>>>> <gerhard.hellriegel@T-ONLINE.DE> wrote:
>>>>
>>>> >On Fri, 29 Feb 2008 09:27:46 +0100, Stefan Pohl
>>>> <stefan.pohl@ISH.DE> wrote:
>>>> >
>>>> >>Dear sas-group,
>>>> >>
>>>> >>I want to read out the information in a string IDs who
>>>> looks like this
>>>> >>
>>>> >>IDs
>>>> >>obs1: [1 2] [3 3] [4 4] [5 5] [6 6] [7 7] [8 8] [9 9]
>>>> >>obs2: [1 4] [5 5] [6 6] [7 7] [8 8] [9 9]
>>>> >>obs3: [1 4] [5 5] [6 8] [9 9]
>>>> >>obs4: [1 9]
>>>> >>
>>>> >>One first idea for a macro readout:
>>>> >>
>>>> >>%MACRO readout(data);
>>>> >>
>>>> >>*obs1;
>>>> >>
>>>> >>%DO i=1 TO 8;
>>>> >>
>>>> >> DATA readout1;
>>>> >>  SET &data;
>>>> >>  WHERE ID in (1,2);  *here the string IDs has to be read out; RUN;
>>>> >>
>>>> >> ....
>>>> >>
>>>> >>%END;
>>>> >>
>>>> >>%MEND readout;
>>>> >>
>>>> >>How can I read out the information in the [...]?
>>>> >>
>>>> >>Best regards, Stefan.
>>>> >
>>>> >
>>>> >Hi Stefan,
>>>> >first I can say, that a macro in that environment might not a good
>>>> >idea,
>>>> >second: mixing macro- and datastep code is a very bad idea!
>>>> >
>>>> >%do i=1 to 8;
>>>> >
>>>> >will never do what you want, whatever it is.
>>>> >
>>>> >One question: what are that "IDs"? What do you want to get?
>>>> >Maybe it would be easier to understand, if you provide the desired
>>>> >result, instead of the method you try to get it.
>>>> >
>>>> >Gerhard
>>>>
>>>> __________ NOD32 2911 (20080229) Information __________
>>>>
>>>> This message was checked by NOD32 antivirus system.
>>>> http://www.eset.com
>>>>
>>>>
>
>
>if those [] are to be ignored, add them among the default delimiters...
>
>Why then is the solution more complex than :>>>>>>>>>
>data have1;
>   infile cards dlm='[ ]' flowover ;
>   input id1 id2 @@ ;
>cards;
>[1 226] [227 390] [391 419] [420 435] [436 439]
>[1 226] [227 390] [391 435] [436 439]
>[1 226] [227 390] [391 439]
>[1 390] [391 439]
>[1 439]
>;
>
>Because the DSD option is not used, consecutive delimiters are treated as
>one.
>I stated the (default) option FLOWOVER, to emphasise that that is the
>required behaviour.
>
>Have I missed some requirement ?
>
>PeterC

I believe the original post specified that a SAS data set with one long
character variable was a given. The DATA HAVE step just modeled that for
testing. Of course one could write it all out to a temporary external file,
then use your solution. And of course, there is always _INFILE_ Magic.