/rsession

R sessions wrapping for Java

Primary LanguageJavaScriptBSD 2-Clause "Simplified" LicenseBSD-2-Clause

Linux & MacOS: Build Status Windows: AppVeyor Build Status

codecov Maven Central

Rsession: R sessions wrapping for Java

Rsession provides an easy to use java class giving access to remote or local R session. The back-end engine is Rserve (locally spawned automatically if necessary, fully compatible with legacy R), Renjin (lower compatibility, but still very good with graphics), and R2js (on-the-fly translation to math.js, lowest compatibility and hack-style coding, full BSD licence). Rsession differs from R2js, Rserve or Renjin as it is a higher level API, and it includes server side startup of Rserve. Therefore, it is easier to use in some point of view, as it provides a multi session R engine (including for Windows, thanks to an ugly turn-around).

Other alternatives:

  • JRI, but it does not provide multi-sessions feature. If you just need one R session in your java code (or work with R environments), JRI is a good solution.

Example Java code

import static org.math.R.*;
...
 
    public static void main(String args[]) {
        Rsession r = RserveSession.newInstanceTry(System.out, null);

        double[] rand = (double[]) r.eval("rnorm(10)"); //create java variable from R command

        //...
        r.set("c", Math.random()); //create R variable from java one

        r.save(new File("save.Rdata"), "c"); //save variables in .Rdata
        r.rm("c"); //delete variable in R environment
        r.load(new File("save.Rdata")); //load R variable from .Rdata

        //...
        r.set("df", new double[][]{{1, 2, 3}, {4, 5, 6}, {7, 8, 9}, {10, 11, 12}}, "x1", "x2", "x3"); //create data frame from given vectors
        double value = (double) (r.eval("df$x1[3]")); //access one value in data frame

        //...
        r.toJPEG(new File("plot.jpg"), 400, 400, "plot(rnorm(10))"); //create jpeg file from R graphical command (like plot)

        String html = r.asHTML("summary(rnorm(100))"); //format in html using R2HTML
        System.out.println(html);

        String txt = r.asString("summary(rnorm(100))"); //format in text
        System.out.println(txt);

        //...
        System.out.println(r.installPackage("sensitivity", true)); //install and load R package
        System.out.println(r.installPackage("DiceKriging", true));

        r.end();
    }

Use it

Using R2js backend:

No dependency required. Only based on Nashorn engine provided in Java >8, so just include https://github.com/yannrichet/rsession/blob/master/Rsession/dist/rsession.jar (latest commit version) in your classpath.

Using Renjin backend:

Add lib/rsession.jar:lib/renjin*.jar in your project classpath:

Using Rserve backend:

Install R from http://cran.r-project.org.

Add lib/rsession.jar:lib/Rserve*.jar:lib/REngine*.jar in your project classpath:

Through maven dependency:

<dependencies>
...
    <dependency>
      <groupId>com.github.yannrichet</groupId>
      <artifactId>Rsession</artifactId>
      <version>3.0.4</version>
    </dependency>
...
</dependencies>

Then, use it in your code (for Windows XP, Mac OS X, Linux 32 & 64):

  • create new Rsession:
    • Renjin (pure Java, no R install necessary):
      Rsession r = new RenjinSession(System.out,null);
    • Rserve (R install required on server side):
      • connect to local Rserve (previously started on localhost with /usr/bin/R CMD Rserve --vanilla --RS-conf Rserve.conf):
        Rsession r = RserveSession.newLocalInstance(System.out,null); 
      • OR connect to local auto-spawned Rserve:
        Rsession r = RserveSession.newInstanceTry(System.out,null);
      • OR connect to remote Rserve (previously started on 192.168.1.1 with /usr/bin/R CMD Rserve --vanilla --RS-conf Rserve.conf):
        Rsession r = RserveSession.newRemoteInstance(System.out,RserverConf.parse("R://192.168.1.1"));
    • R2js:
      Rsession r = new R2jsSession(System.out,null);
  • do your work in R and get Java objects
    • create Java objects from R command using
    HashMap<String,Object> vars = ...
    vars.put("a",42);
    Object o = r.eval("pi+a",vars);
    (Object o is automatically cast to double, double[], double[][],String, String[], ...)
    • OR
      • create your R objects using r.set("a",42)
      • call any R command using r.eval("pi+a")
      • if needed use R packages install & load: r.installPackage("MASS", true);
      • you can access R command answers as string using: r.asString("...") , r.toPNG(File f,"plot(rnorm(10))")
  • finally close your Rsession instance: r.end();

Analytics