Date:         Fri, 5 Feb 1999 14:09:39 GMT
Reply-To:     alte@RZ.UNI-GREIFSWALD.DE
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Shiling Zhang <shiling@MATH.WAYNE.EDU>
Organization: Deja News - The Leader in Internet Discussion
Subject:      Re: QUESTION: reading unorthodox data from flat file

Dietrich,
 
You can use a line holder @.
 
1) read the first field.
2) if the first field is a count number, then read the 2nd and 3rd fields,
else read the whole line as a comment field.
 
3) combine those wrapped comments into one.
 
Hope this helps.
 
data t1;
  retain id case;
  length id 8 case $8 comments $100 testid $15;
  infile '~/test.txt'  pad;
  input testid @;
    if not indexc(upcase(testid),'ABCDEFGHIJKLMNOPQRSTUVWXYZ') then do;
      id=input(testid, 6.);
      input case comments ;
     end;
    else do;
     input comments 1-100;
    end;
 
keep id case comments;
run;
 
*combine comments;
proc sort;by id;run;
 
 data t2;
 retain id case cmtsall;
 length cmtsall $200;
   set t1;
    by id;
    if first.id then do;
     cmtsall='';
    end;
    cmtsall=compbl(cmtsall||comments);
    if last.id then do;
     put id= case= cmtsall=;
     output;
    end;
 drop comments;
 run;
****************************
Subject: QUESTION: reading unorthodox data from flat file
Date: Thu, 4 Feb 1999 16:42:03 -0000
From: alte <alte@RZ.UNI-GREIFSWALD.DE>
Organization: CM
To: SAS-L@UGA.CC.UGA.EDU
Newsgroups: bit.listserv.sas-l
 
 
hi,
 
i am trying to read ascii data into sas (6.12 NT). it looks like this
 
001     111112 Mein83_1         some text
002     213455 Mein83_2         other text
003     135462 Mein83_3         more text
004     351842 BESOND           complications
005     no money
006     critical value
007     undetermined origin
008     321682 BESOND           text
009     321654 Wohn9we          Buspark
 
in a *standard* case the first 6 chars are an ID for the case, the second
collection of chars is the variable name and after some empty spaces there
comes some text.
when the data typist enters line feeds, then our data entry tool (Ci3 by
sawtooth)
gives wrapped lines in the flat file without adding the case ID to them in
the beginning.
so the line text lines 005 to 007 belong to case 351842 (line 004). i don't
know before when wrapped lines occur, neither how many there will be.
 
i would like to have the case ID in one variable and the variable name in
another and the complete text for one case in a 3rd, e.g. ID VARNAME TEXT.
 
how could that be done?
 
any help would help ;-)
 
d. alte
 
-------------------------------------------------------------
  Dietrich Alte, Diplom-Statistiker
 
  Universitaet Greifswald
  Institut fuer Epidemiologie und Sozialmedizin
  Organisationszentrum Community Medicine
  Walther-Rathenau-Str. 48
  17487 Greifswald
 
  phone +49 (0) 3834 - 86 77 13
  fax     +49 (0) 3834 - 86 66 84
 
 email alte@rz.uni-greifswald.de
-------------------------------------------------------------