[scilab-Users] Using functions in Java
Sylvestre Ledru
sylvestre.ledru at scilab.org
Tue Jun 16 09:35:21 CEST 2009
Le lundi 15 juin 2009 à 21:34 -0700, Sattiraju Prabhakar a écrit :
> Thanks Sylvestre! Your response is very helpful.
Good
> Sorry for the delayed reply.
No worries
> First, let me explain why I was trying to create Java methods for
> Scilab functions, without using Scilab.ExecuteScilabScript(String filename).
> I have been developing a complex Java application, and a lot of effort
> has gone into structuring the application to make it extensible and
> readable. I would like Java to continue to provide that structure while
> using other programs like Scilab. For example, I can have variations of
> methods as the classes get extended.
>
> I agree with you that using Scilab.ExecuteScilabScript(String filename)
> saves time.
>
> I tested your suggestion for "Problem Using Function" below -include "end".
> It is still giving error. When I tried to step through the function
> declaration, in debugger, I found something interesting.
> Everytime Scilab.Exec() encountered, it returns "false",
> except in the case of "disp". I think Scilab.Exec requires an
> executable statements as arguments.
Yep, it is what I tried to say in my previous email.
> Since the function declaration does not provide statements for immediate execution, it returns false
> (error, according to Scialb.Exec()). I think that is also the
> reason why it executes disp statement during the function declaration.
> Am I correct? Is there a way around this problem, without entering the function
> body as a very long string?
It is not really a problem. Each instruction has to be syntactically correct...
If you do want to be able to do that, you can inherent from the Scilab
class and create a SQL-commit like feature.
Ie, write into a buffer your series of commands and once you are happy,
you commit it (and the commit with just merge all the strings into one
or write into a temp file and Scilab.ExecuteScilabScript it).
Am I clear ?
> Thanks for suggesting me to include "end". The "string" type solution which you provided
> is working for me too.
Good.
> I guess the same problems are applicable to "deff" also.
Probably !
Sylvestre
> Thanks!
>
> Prabhakar
>
>
>
>
> > -----Original Message-----
> > From: Sylvestre Ledru [mailto:sylvestre.ledru at scilab.org]
> > Sent: Wednesday, June 10, 2009 8:47 AM
> > To: users at lists.scilab.org
> > Cc: Sattiraju Prabhakar
> > Subject: Re: [scilab-Users] Using functions in Java
> >
> > Hello,
> >
> >
> > Le samedi 30 mai 2009 à 01:15 -0700, Sattiraju Prabhakar a écrit :
> > > Hi,
> > >
> > > I am trying to use Scilab (5.1.1) from a Java (jdk 6 update 13)
> > > application.
> > > I have been experimenting with some Scilab scripts that use
> > > function ("function" and "deff") definitions.
> > > When I tried to convert these scripts into Java, I have come
> > > across a number of problems.
> > >
> > > Problems using "function"
> > > ================
> > >
> > > I took the "function percentile" declaration from Professor Gilberto
> > > Gurroz's
> > > Basic Statistics and Probability using Scilab. I reproduce below the
> > > script.
> > >
> > > function [p] = percentile(x,r)
> > > //This function calculates the 100*r-th percentile
> > > //(0<r<1) for the vector x
> > > xx = gsort(x);
> > > [n m] = size(xx);
> > > if m>n & n == 1
> > > n = m;
> > > end
> > > if r<0 | r>1 then
> > > disp('Percentile value must be between 0 and 1');
> > > else
> > > k = n*r;
> > > if k-floor(k) ~= 0
> > > p = xx(round(n*r));
> > > else
> > > p = (xx(k)+xx(k+1))/2;
> > > end
> > > end
> > >
> > > It is used as follows
> > > Q1 = percentile(x,0.25)
> > >
> > > I tried to put it in Java code in two different ways.
> > >
> > > A. In the first attempt, I tried to execute each line using
> > > Scilab.Exec. The code follows.
> > >
> > > public void percentile() {
> > > Scilab.Exec("function [p] = percentile(x,r)");
> > > //This function calculates the 100*r-th percentile
> > > //(0<r<1) for the vector x
> > > Scilab.Exec("xx = gsort(x);");
> > > Scilab.Exec("[n m] = size(xx);");
> > > Scilab.Exec("if m>n & n == 1");
> > > Scilab.Exec("n = m;");
> > > Scilab.Exec("end");
> > > Scilab.Exec("if r<0 | r>1 then");
> > > Scilab.Exec("disp('Percentile value must be between 0 and
> > > 1');");
> > > Scilab.Exec("else");
> > > Scilab.Exec("k = n*r;");
> > > Scilab.Exec(" if k-floor(k) ~= 0");
> > > Scilab.Exec("p = xx(round(n*r)); ");
> > > Scilab.Exec("else");
> > > Scilab.Exec("p = (xx(k)+xx(k+1))/2;");
> > > Scilab.Exec("end");
> > > Scilab.Exec("endfunction");
> > > }
> > Why don't you directly store this into a dedicated percentile.sci file
> > and call the java method Scilab.ExecuteScilabScript(String filename) ?
> >
> > Btw, your error is due to a missing end just before the endfunction.
> >
> >
> > > I tried to apply this function as follows.
> > > SciDouble Q1P = new SciDouble("Q1");
> > > Scilab.Exec("Q1 = percentile(x,0.25)");
> > > Q1P.Get();
> > > Q1P.disp();
> > >
> > > I get the following error.
> > > double Q1=
> > >
> > > 0.
> > > Error in Java_javasci_Scilab_Exec routine.
> > This is normal that it is failling here.
> > Scilab.Exec is waiting for a consistent expression.
> >
> > If you try:
> > if (!Scilab.Exec("function [p] = percentile(x,r)")){
> > System.out.println(Scilab.GetLastErrorCode());
> > }
> > you will that that you will get an error (code 272... no endfunction).
> >
> >
> > > B. I tried to declare the entire function as a string and pass the
> > > string as argument to Scilab.Exec().
> > >
> > >
> > > public void percentile() {
> > > String perc2 = "function [p] = percentile(x,r) \n xx = gsort(x);
> > > \n [n m] = size(xx); \n if m>n & n == 1 \n n = m; \n end \n if r<0 |
> > > r>1 then \n disp('Percentile value must be between 0 and 1'); \n else
> > > \n k = n*r; \n if k-floor(k) ~= 0 \n p = xx(round(n*r)); \n else \n p
> > > = (xx(k)+xx(k+1))/2; \n end \n endfunction";
> > > Scilab.Exec(perc2);
> > > }
> >
> > > I tried it without \n, and later with \n. In both cases, I got the
> > > same error as above.
> > Same issue, you forgot the end.
> >
> > The following code is working for me:
> > import javasci.*;
> > public class percentile {
> >
> >
> > public static void main(String args[]){
> > String aze="function [p] = percentile(x,r); xx = gsort(x);
> > [n m] =
> > size(xx); if m>n & n == 1 n = m; end; if r<0
> > | r>1 then disp('Percentile value must be between 0 and 1');
> > else; k = n*r; if k-floor(k) ~= 0; p =
> > xx(round(n*r)); else; p = (xx(k)+xx(k+1))/2; end;
> > end; endfunction";
> > if (!Scilab.Exec(aze)){
> > System.out.println(Scilab.GetLastErrorCode());
> > }
> > SciDouble Q1P = new SciDouble("Q1");
> > Scilab.Exec("x=[23,2323,934,42];");
> > Scilab.Exec("Q1 = percentile(x,0.25)");
> > Q1P.Get();
> > Q1P.disp();
> > }
> > }
> >
> > > 2. Using deff
> > > =============
> > >
> > > I was able to successfully write and execute the above example using
> > deff
> > >
> > > Scilab.Exec("deff('[x,y]=myfct(a,b)', ['x=a+b'; 'y=a-b'])");
> > >
> > > "deff" requires that we state the "statements" of the function as
> > matrix of strings.
> > > If you have a String inside the function declaration (of Scilab
> > script), as in percentile, then
> > > it becomes a problem.
> > >
> > > I was not able to get a correct result for percentile using deff and
> > Java.
> > > Here is my code:
> > > Scilab.Exec("deff('[p]=percentile(x,r)', ['xx = gsort(x);'; '[n m] =
> > size(xx);'; ' if m>n & n == 1 n = m;'; 'end'; 'if r<0 | r>1 then
> > disp('Percentile value must be between 0 and 1');' ; 'else k = n*r;';
> > 'if k-floor(k) ~= 0 p = xx(round(n*r));'; 'else p =
> > (xx(k)+xx(k+1))/2;'; 'end' ])");
> > Probably the same issue as before.
> >
> > > I would like to end with a general question. Each of the above methods
> > > takes some effort.
> > > If the function definition is long, then the effort needed to convert
> > > into Java code can be
> > > substantial. Is there a simple method?
> > Yep, load the function from a file.
> >
> > Sylvestre
>
>
>
More information about the users
mailing list