Date: Fri, 10 Mar 2000 14:43:08 +0100
Reply-To: HALIMI Raphaël <Raphael.HALIMI@GEP.FR>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: HALIMI Raphaël <Raphael.HALIMI@GEP.FR>
Subject: Re: text file -> SAS data set
Content-Type: text/plain; charset="iso-8859-1"
Jim,
I always enjoy reading your solutions. At each times, I learn something !
By the way, I've tried this solution :
data read;
infile myfile;
input FullLine $;
run;
And the results :
OBS READ
--------------
1 IF
2 THEN
3 ELSE
4 IF
5 THEN
. ELSE
. .
. .
. .
So I tried :
data read;
infile myfile delimiter="]";
input FullLine $;
run;
And the results :
OBS READ
-----------------
1 IF X4 = "1"
2 THEN Y = "0" [49795 0
3 ELSE
4 IF X3 = "3"
5 THEN Y = "0" [12604 0
6 ELSE
7 IF X3 = "1"
8 THEN Y = "0" [12474 0
9 ELSE
. .
. .
. .
---> That's why I was looking for several delimiters (that is : IF, THEN,
and ELSE). Now, I can't understand why the 2nd of my programs isn't correct
!? I thought I will have this kind of data :
OBS READ
-----------------
1 IF X4 = "1" THEN Y = "0" [49795 0
2 ELSE IF X3 = "3" THEN Y = "0" [12604 0
3 ELSE IF X3 = "1" THEN Y = "0" [12474 0
..........
Do I make an obvious mistake ?
> ----------
> De : Jim Groeneveld[SMTP:J.Groeneveld@ITGROUPS.COM]
> Répondre à : Jim Groeneveld
> Date : vendredi 10 mars 2000 12:41
> A : SAS-L@LISTSERV.UGA.EDU
> Objet : Re: text file -> SAS data set
>
> Raphaël,
>
> Regarding your first issue, I would ask whether it is possible (and
> possibly
> more easy) to force your CN program to deliver the desired values more
> suitedly. If that is not possible I would read and analyse your file
> approximately in the following *untested* way:
> 1. read a line (a line as in your message) into the variable FullLine
> (after
> defining your INFILE)
> INPUT FullLine $;
> 2. if it starts with IF the remainder is the variable Cond
> IF UPCASE(SUBSTR(FullLine,1,2)) EQ "IF" THEN
> Cond=SUBSTR(FullLine,3);
> 3. if it starts with ELSE, ignore, don't process further
> ELSE IF UPCASE(SUBSTR(FullLine,1,4)) EQ "ELSE" THEN ;
> 4. if it starts with THEN the remainder up to the [ is the variable Conclu
> ELSE IF UPCASE(SUBSTR(FullLine,1,4)) EQ "THEN" THEN
> DO;
> Bracket = INDEX(FullLine,'['); * assumed always to be
> positively found;
> Conclu = SUBSTR(FullLine, 5,Bracket-5);
> from the [ up to the end it is the variable Size
> Size = SUBSTR(FullLine,Bracket);
> OUTPUT; * force an observation written
> after
> found Size;
> END;
> 5. in any other case it is an additional part of the Cond
> ELSE Cond=Cond||FullLine;
> Assuming I did interprete your file correctly and this is the actual
> layout.
> Process starting, embedded and trailing spaces within the lines and
> variables according to your own desire.
>
> To answer your additional question in attempt 1: yes it it possible to
> define more than one delimiter at the same time, but shouldn't you include
> a
> space as well?
> To answer your additional question in attempt 2: an INPUT statement like
> <input @"IF" cond @"THEN" conclu @"["> is legal (and I do see how you
> would
> like to have it work), but as you don't know which literal character
> string
> to expect at any moment this method is not useful, but you should follow
> some way like outlined above, reading and analysing complete lines.
>
> Your suggestion to read a complete text file into a macro variable IMHO
> will
> not work, as the %INCLUDE statement is not a macro statement at all
> (resolving to the contents of the file); it is an exception.
>
> A way to read a complete textfile into a single macro variable (FullFile)
> is
> *untested, but the method has been applied in some other situation*:
> %LET FullFile = ; * Define empty macro variable at this level;
> DATA dataset;
> INFILE <specification(s)>;
> LENGTH FullLine $ 80; * or longer;
> INPUT FullLine $;
> CALL EXECUTE ( '%ValueLst(FullFile,' || FullLine || ');' );
> RUN;
>
> where ValueLst is the following *tested*, preceding macro:
>
> %*~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~;
> %* Auxiliary ValueLst concatenates values into list ;
> %*__________________________________________________;
>
> %MACRO ValueLst(ListVar,Value,ForMyUse) / DES='Make value list from
> values'
> /*STORE*/;
> %IF ( %UPCASE(&ForMyUse) = COMPILE ) %THEN %GOTO Finish;
> %LET &ListVar = &&&ListVar &Value; %* &ListVar already exist in calling
> macro;
> %Finish:
> %MEND ValueLst;
>
> Ignore or remove the parts refering ForMyUse; I use those for compilation
> and other author-only purposes. After the data step your whole text file
> is
> available in the macro variable FullFile, to be used subsequently. You
> might
> use this method to deduce the contents of your variables alternatively
> (not
> suggested here).
>
> Regards - Jim.
> --
> Y. (Jim) Groeneveld, MSc IMRO TRAMARKO tel. +31 412 407 070
> senior statistician, P.O. Box 1 fax. +31 412 407 080
> head IT department 5350 AA BERGHEM IMRO TRAMARKO: a CRO
> J.Groeneveld@ITGroups.com the Netherlands in clinical research
>
> My computer seems more ²°°°-compatible than I am myself.
>
>
> > -----Original Message-----
> > From: HALIMI Raphaël [SMTP:Raphael.HALIMI@GEP.FR]
> > Sent: Friday, March 10, 2000 11:18 AM
> > To: SAS-L@LISTSERV.UGA.EDU
> > Subject: text file -> SAS data set
> >
> > Dear SAS-lers,
> > This the new episode of "SAS & CN algorithm" (CN is a rule-induction
> > algorithm).
> > Here is the CN output, in a text file :
> >
> > IF X4 = "1"
> > THEN Y = "0" [49795 0]
> > ELSE
> > IF X3 = "3"
> > THEN Y = "0" [12604 0]
> > ELSE
> > IF X5 > 4.50
> > AND X6 > 2.50
> > THEN Y = "0" [390 0]
> > ELSE
> > IF X1 > 64.96
> > THEN Y = "1" [0 557]
> >
> > What I'd like to do is to transform this text file in a data set :
> >
> > COND conclu size
> >
> --------------------------------------------------------------------------
> > --
> > ----------
> > X4 = "1" Y = "0" [49795 0]
> > X3 = "3" Y= "0" [12604 0]
> > X5 > 4.50 AND X6 > 2.50 Y = "0" [390 0]
> > X1 > 64.96 Y= "1" [0 557]
> >
> > What I try (and fail) to do was :
> > 1- working with several delimiters with an INFILE statement : impossible
> ?
> > (delimiter="IFTHEN[")
> > 2- working with an INPUT statement : input @"IF" cond @"THEN" conclu
> @"["
> > size;
> >
> > Any help would be welcome,
> > Thanks everyone !
> >
> > PS : is there an easy way to assign a macro-variable a whole text file
> > (something like : %LET read= %STR(%include "path/text.txt";); ???)
> >
> > Raphael
>
| 
